A systematic search was conducted across the PubMed, Web of Science, Embase, Cochrane Library, CNKI, Wanfang Data, and VIP databases for studies investigating AS-IV intervention in rodent models of myocardial fibrosis. The search spanned from each database’s inception through 24 November 2024. Key search terms included: 1) Astragaloside IV/AS-IV; 2) myocardial fibrosis; 3) rats/mice. Terms were combined using logical “AND” operators. The specific search formula is provided in Supplementary Material 1.

1) Study subjects: Rodent models with MF confirmed by indicator detection.

1) The data for evaluation indicators were incomplete.

According to the method of including studies in version 5.0.2 of the Cochrane Collaboration’s Handbook for Systematic Reviewers, the retrieval results from each database were imported into the literature management software Zotero. In parallel and independently, two reviewers, Li Haozhe and Chu Yunhang, screened the literature and then extracted the data and cross-checked the experimental results. In case of disagreement, they reached a decision through negotiation or through referring the matter to Wang Yue for judgment. Finally, the eligible literature data were classified and statistically analyzed using MS Excel. The data to be extracted included author information, publication year of the literature, drugs used in the experiment, specific modeling methods, number of models, route of drug administration, numerical value of the drug dosage, categories of detection indicators, corresponding units of the detection indicators, and final experimental result data. To ensure the validity of the data and the reliability of the analysis, when the number of studies related to a certain outcome indicator was less than 3, this outcome indicator would be excluded from the scope of the study.

Additionally, this systematic review was registered in the PROSPERO International Prospective Register of Systematic Reviews (registration number: CRD420250637182) to help avoid duplication and reduce the possibility of reporting bias by comparing the completed evaluation with the planned protocol.

Meta-analyses were performed using RevMan 5.1 software. For categorical data, the risk ratio (RR) was used as the effect size measure. For continuous data, the mean difference (MD) was employed. When outcome measures shared identical units and methods, the weighted mean difference (WMD) was calculated; otherwise, the standardized mean difference (SMD) was used. All effect sizes were reported with 95% confidence intervals (CI). Heterogeneity among included studies was evaluated using Cochran’s Q test (significance level: P < 0.1) and quantified by the I² statistic. A fixed-effect model was applied when I² ≤ 50% indicated low heterogeneity, while a random-effect model was used for high heterogeneity (I² > 50%), followed by exploration of heterogeneity sources. Publication bias was initially assessed visually via funnel plots. Additionally, StataSE 12.0 software was used to conduct a leave-one-out method to investigate potential heterogeneity sources, complemented by Egger’s test for publication bias detection. Subgroup analyses and meta-regression analyses were performed to identify the root causes of significant heterogeneity.

The literature retrieval process for this meta-analysis is outlined in Figure 2. A total of 616 relevant articles were identified through searches in databases including PubMed. After systematic screening, 38 eligible studies see Table 1 for details. (Rios et al., 2022; Chen and Wang, 2017; Fathiazad et al., 2019; Jiang and Zhang, 2016; Jiang and Wei, 2016; Jiang, 2016; Lin et al., 2019; Luo et al., 2020; Lv, 2018; Liu et al., 2022; Li et al., 2005; Li, 2021; Chen et al., 2011; Tang et al., 2016; Tang, 2017; Tang et al., 2017a; Liu et al., 2019; Wu et al., 2023; Wang SF. et al., 2020; Zhang et al., 2022; Xue et al., 2024; Xu et al., 2010; Zhao, 2020; Zhang et al., 2007; Zhang, 2021; Wang Z. et al., 2020; Cheng, 2017; Gong and Ke, 2024; He, 2013; Li et al., 2017; Li, 2013; Tang et al., 2017b; Wang et al., 2024; Wang XL. et al., 2021; Yan et al., 2024; Zhang, 2012; Zhang, 2015; Zhang et al., 2023) were included in the final analysis. Data extraction included study-specific identifiers, experimental subjects, modeling methods, and other study characteristics. Among the included studies, those with three groups typically consisted of a sham operation group, a model group, and an astragaloside IV treatment group. Studies with five groups additionally incorporated three subgroups receiving different doses of AS-IV, in addition to the sham and model groups.

This meta-analysis included 38 designs involving 1,334 rodent models (891 in AS-IV treatment groups vs. 443 in control groups), with species distribution as follows: 25 studies used Sprague–Dawley rats (Rios et al., 2022; Fathiazad et al., 2019; Jiang and Zhang, 2016; Jiang and Wei, 2016; Lin et al., 2019; Luo et al., 2020; Li et al., 2005; Chen et al., 2011; Tang et al., 2016; Tang, 2017; Wu et al., 2023; Zhang et al., 2022; Xu et al., 2010; Zhang, 2021; Wang Z. et al., 2020; Cheng, 2017; Gong and Ke, 2024; Li et al., 2017; Li, 2013; Tang et al., 2017b; Wang XL. et al., 2021; Yan et al., 2024; Zhang, 2012; Zhang, 2015; Zhang et al., 2023), three studies used Wistar rats (Chen and Wang, 2017; Lv, 2018; Liu et al., 2022), one study used Dahl SS rats (Zhang et al., 2007), five studies used C57BL/6J mice (Jiang, 2016; Tang et al., 2017a; Liu et al., 2019; Wang SF. et al., 2020; He, 2013), two studies used BALB/c mice (Li, 2021; Zhao, 2020), and one study each used Kunming mice (Xue et al., 2024) and ICR mice (Wang et al., 2024).

Quality assessment results: The included studies were evaluated using SYRCLE’s Risk of Bias tool for animal experiments. Among all included studies, 25 clearly stated that they used the random number table method to generate allocation sequences (Chen and Wang, 2017; Fathiazad et al., 2019; Jiang and Zhang, 2016; Jiang, 2016; Lin et al., 2019; Luo et al., 2020; Lv, 2018; Liu et al., 2022; Li et al., 2005; Chen et al., 2011; Tang et al., 2016; Tang, 2017; Tang et al., 2017a; Liu et al., 2019; Wu et al., 2023; Wang SF. et al., 2020; Zhang et al., 2022; Xue et al., 2024; Xu et al., 2010; Zhao, 2020; Zhang et al., 2007; Zhang, 2021; Wang Z. et al., 2020; Gong and Ke, 2024; He, 2013; Li et al., 2017; Li, 2013; Tang et al., 2017b; Wang et al., 2024; Wang XL. et al., 2021; Zhang, 2012; Zhang, 2015; Zhang et al., 2023; Jia et al., 2023), and two did not adopt this method for determining allocation sequences (Jiang and Wei, 2016; Li, 2021). Studies that did not adopt the random number table method may introduce a certain degree of subjectivity into the grouping process, making it difficult to ensure fairness and comparability between groups, thereby affecting the reliability of the research results. Two studies provided no clear description of how they allocated subjects (Cheng, 2017; Yan et al., 2024), and the scientificity and rationality of their allocation sequence generation are questionable, which increases the risk of bias. However, none of the studies mentioned allocation concealment, random housing of animals, or blinding of relevant personnel and outcome assessors. The absence of these key links may introduce biases throughout the process from grouping and feeding to outcome assessment, ultimately undermining the credibility of the experimental conclusions regarding astragaloside IV. All included animals were incorporated into the final analysis, with no instances of selectively choosing animals for evaluation. Although three studies contained incomplete data (Chen and Wang, 2017; Lin et al., 2019; Liu et al., 2019), these gaps were accurately explained and reasonably justified, confirming the completeness of reporting and ruling out any association with selective reporting see Table 2 for details.

AS-IV significantly influenced cardiac morphological parameters (CVF, LVESd, LVEDd, IVSd, LVPWd, and LVMI), hemodynamic indices (LVFS, LVEDp, LVSP, LVEF, and ±dp/dtmax), and biochemical/physiological indicators (LDH, TNF-α, BW, and HR) in rodent models of cardiac disease. However, it exerted no significant statistical effect on LVIDd and LVIDs.

In the present study, meta-regression analyses were conducted on indicators exhibiting high statistical heterogeneity with 10 or more included studies. Using species type, administration dose, and administration duration as covariates, these analyses aimed to explore potential sources of heterogeneity. The results demonstrated that species-related factors exerted a significant regulatory effect on LVPWd and CVF (P < 0.05); administration dose significantly influenced the effect sizes of LVEDd, LVEDp, LVPWd, and CVF (P < 0.05); and administration duration, in turn, exerted a significant regulatory influence on the effect sizes of LVFS, LVEDp, LVPWd, and +dp/dtmax (P < 0.05). Species, dose, and duration, operating through distinct mechanisms, emerged as significant contributors to the heterogeneity observed in the aforementioned indicators. Meanwhile, indicators such as IVSd, LVIDs, and body weight, due to having fewer than 10 included studies, lacked adequate data support. The sources of their heterogeneity require subsequent analysis with an expanded sample size.

Additionally, given the significant heterogeneity in studies examining astragaloside IV’s effects on outcome indicators, we conducted concurrent subgroup analyses based on the basic characteristics of the included literature. These analyses aimed to further explore, validate, and refine the conclusions drawn from the meta-regression.

CVF stands as a key histological metric for gauging MF severity, offering a quantitative readout of fibrotic burden by assessing collagen content in myocardial tissue. Higher CVF values signify greater collagen accumulation and more marked fibrotic features (Lang et al., 2025). The current study showed that AS-IV notably lowers CVF, pointing to its potential to alleviate MF. Moreover, these findings suggest CVF regulation in MF ties to the TGF-β/Smad signaling pathway (Xu et al., 2023).

In rat models with MF, CVF was significantly elevated alongside upregulated expression of TGF-β1, phosphorylated Smad2 (P-Smad2), and type I collagen—changes that in turn promote collagen synthesis and deposition (Zeng et al., 2023). These results mirror those from hypoxia-induced myocardial hypertrophy models, where AS-IV reduces collagen deposition (particularly type I collagen) by inhibiting the TGF-β1/Smad2 pathway, lowers CVF, and also increases IVSd and LVMI (Zhang et al., 2024).

Meanwhile, increases in LVESd and LVEDd raise myocardial wall tension, in turn exacerbating myocardial injury and fibrosis. Reducing LVESd and LVEDd thus plays a key role in improving ventricular perfusion. AS-IV has been shown to mitigate ventricular dilation by repairing myocardial damage, boosting contractile function, reducing lipid accumulation, and downregulating pro-fibrotic factor expression (Li et al., 2023a). Additionally, its derivative HHQ16 acts directly on cardiomyocytes, helping reverse myocardial hypertrophy and post-infarction ventricular remodeling, reducing LVESd and LVEDd, and significantly improving cardiac function (Wan et al., 2023).

Taken together, AS-IV regulates CVF, lowers such markers as LVESd, LVEDd, IVSd, and LVMI, and inhibits activation of the TGF-β/Smad pathway—effects that in turn reduce collagen buildup and effectively curb MF.

TNF-α is a core pro-inflammatory mediator, and AS-IV can improve its level. It may alleviate 2-deoxy-D-glucose (2-DG)-induced apoptosis in PC12 cells by inhibiting endoplasmic reticulum stress and blocking the opening of the mitochondrial permeability transition pore (mPTP), thereby significantly reducing the expression of the apoptotic marker Caspase-3 (Wan et al., 2023) and further mitigating the inflammatory response. Furthermore, it can promote M2 microglial polarization by activating the PI3K/AKT signaling pathway, thereby alleviating neuroinflammation and cerebral damage (Wang Z. et al., 2024). These findings suggest that AS-IV may exert cardioprotective effects through similar molecular mechanisms.

In a lipopolysaccharide (LPS)-induced cardiac dysfunction model, AS-IV dampens NLRP3 inflammasome activation, reducing the release of pro-inflammatory cytokines like IL-1β. At the same time, it reins in mitochondrial function and curbs the production of reactive oxygen species (ROS) (Fu et al., 2020; Feng et al., 2021). It also increases the PPARγ signaling pathway, strengthens the Nrf2/HO-1 antioxidant pathway, and reduces excessive iNOS and nitric oxide production (Liang et al., 2023).

These mechanisms point to AS-IV’s ability to exert a synergistic regulatory effect on cardiomyocyte apoptosis and inflammatory responses via multi-target actions. Its impacts span antioxidation, mitochondrial protection, and metabolic reprogramming—all of which lay a theoretical groundwork for its use in cardiovascular diseases (Chen et al., 2021; Zaman et al., 2022; Zhai et al., 2024).

AS-IV reduces LVEDp significantly by boosting myocardial diastolic function. In myocardial ischemia-reperfusion (I/R) injury models, pre-treating with AS-IV reduces microvascular leakage (MVL), which in turn eases cardiac edema and diastolic dysfunction, consistent with the drop in LVEDp (He et al., 2022). At the same time, AS-IV can repair mitochondrial function, curb ROS buildup, and improve calcium handling in cardiomyocytes, all of which lower LVEDp (Wang et al., 2024b; Wang et al., 2024c).

In sepsis-induced cardiomyopathy models, AS-IV increases myocardial contractility by restoring mitochondrial balance and endoplasmic reticulum function (Su et al., 2022). It also increases endothelial nitric oxide synthase (eNOS) activity, spurring more nitric oxide (NO) production. This mitigates oxidative stress-related myocardial damage and keeps LVSP stable (Meng et al., 2021; Leng et al., 2020).

Studies show AS-IV improves ventricular geometric remodeling by shrinking LVEDp and LVSP. In chronic kidney disease (CKD) models with concurrent myocardial injury, AS-IV eases ventricular dilation by blocking the renin-angiotensin system (RAS) and fibrotic signaling pathways (Li et al., 2023b; Li et al., 2022). Notably, combining AS-IV with other active compounds like tanshinone IIA synergistically enhances LVEDp and LVSP, yielding a more robust cardioprotective effect (Zhai et al., 2024).

In short, AS-IV eases cardiac dysfunction by comprehensively regulating LVEDp and LVSP through a multi-target mechanism.

AS-IV, the primary bioactive component in Astragalus, delivers a range of cardioprotective effects by boosting both systolic and diastolic heart function. Under hypoxic conditions, it significantly eases myocardial hypertrophy and cardiac injury—effects likely tied to curbing oxidative stress and regulating energy metabolism (Zhang et al., 2024).

What is more, in diabetic cardiomyopathy (DCM), AS-IV helps repair myocardial damage, increases contractile function, and reduces lipid buildup (Li et al., 2023a). It also enhances vascular relaxation by reversing oxidative stress-induced endothelial dysfunction, doing so by increasing eNOS activity and boosting NO levels.

AS-IV markedly improves cardiac systolic and diastolic function through multi-target, multi-pathway actions. Its key mechanisms include antioxidant and anti-inflammatory activity, regulating mitochondrial dynamics, and tweaking various signaling pathways (He et al., 2022).

Myocardial cells in underweight individuals often suffer from long-term malnutrition, which impairs their energy metabolism and weakens their self-repair capacity (Fernandez-Patron et al., 2024). Studies show AS-IV can boost body weight in rodent models, hinting that the compound might aid myocardial repair and regeneration by ensuring adequate protein supply, thus slowing fibrosis progression.

Weight gain correlates with preserving key cardiac function markers like myocardial contractility and cardiac output, which may ease fibrosis’s harmful impact on heart performance. What is more, a reduced heart rate can improve myocardial oxygen use, reduce cardiac afterload, and, in turn, slow the progression of MF.

This systematic review has certain limitations: 1) Currently, the pharmacokinetic characteristics of astragaloside IV in humans remain unclear, and the efficacy observed in animal experiments cannot be directly extrapolated to clinical dosing regimens in humans; 2) most animal experiments involve short-term interventions, which do not align with the actual clinical scenario of long-term fibrosis requiring continuous treatment. This may lead to an underestimation of long-term drug efficacy or overlook delayed adverse reactions, thereby affecting the benefit-risk assessment; 3) the pathogenic mechanisms of animal models differ significantly from those of human chronic fibrosis, resulting in model-dependent bias; 4) there is uncertainty regarding the application of allocation concealment in all included studies, and it remains unclear whether implementers of interventions used blinding methods, which introduces considerable selection bias into the results; 5) variations in administration dosages and intervention durations are likely to induce heterogeneity in outcomes, undermining the stability, reliability, and generalizability of the conclusions. It is anticipated that more high-quality literature will be included in future studies to obtain more robust evidence-based medical evidence.