Effects of Resveratrol on MCP-1/CCL2-Related Readouts in Preclinical Animal Models: A Systematic Review and Meta-Analysis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Chiu et al. aim to investigate the effects of resveratrol on MCP-1/CCL2-related inflammatory signaling in animal models by systematic review and meta-analysis. To achieve this aim, the authors conducted a PubMed search to identify MCP-1/CCL2-related in vivo studies involving resveratrol treatment. After obtaining the relevant study groups, they conducted a meta-analysis and tried to fit the findings into a meta-regression model. They also performed additional analyses ranging from checking sensitivity to risk-of-bias assessment to the robustness of resveratrol treatment for blocking MCP-1/CCR2 axis chronic inflammation.

Resveratrol, a naturally occurring plant-derived polyphenol, has been widely studied for its broad ability to regulate systemic inflammation and oxidative stress. Despite its pharmacokinetic limitations, identifying the molecular targets of resveratrol is important for understanding its role in chronic inflammation. One key target is the MCP-1/CCL2–CCR2 axis, which regulates monocyte and macrophage recruitment into damaged or inflamed tissues.

Below you can find the points to be considered:

1. The study includes only a PubMed search; as far as I know, at least two databases should be searched.
2. The phrase “MCP-1/CCL2 expression” does not fully match the content of the study. The techniques used in the included datasets differ, including relative RNA levels, ELISA, and optical density. Therefore, instead of “expression,” a phrase such as “MCP-1/CCL2-related readouts” may be more appropriate.
3. Since the data are predominantly from rodent models, the dose meta-regression is non-significant (p = 0.482), heterogeneity is high (I² = 78.9%), and the effect is in the opposite direction in the piglet model, strong claims should be avoided. For example, the phrases “robustly attenuates” in the Discussion section, lines 340–341, and “heavily mitigating monocyte-driven inflammatory cascades” in the Conclusions section, lines 386–387, could be softened.
4. Figure 1: The reasons for excluding the 58 records during title/abstract screening are not provided. Additionally, the criteria used by the “automation tool” that excluded 12 studies before screening is not clear.
5. Figure 2 could be moved to the Supplementary section.
6. In Figure 3, the term “pooled” is used for piglets and rabbits, although there is only one dataset for each.
7. In Supplementary Table S5, data are provided for only 9 studies, whereas Figure 4B and the meta-regression model appear to use the full dataset. The conclusion that dose did not explain heterogeneity (p = 0.482) is difficult to interpret without a complete dose table showing the dose data used in the meta-regression model.
8. The heterogeneity data are inconsistent: I² is reported as 78.9% in the Abstract, line 38, and in the Results section, line 234, whereas it is stated as I² > 80% in the Discussion section, line 374.
9. References [40] and [47] are duplicates of each other; both refer to Cheng et al., BioMed Research International, 2020.

Author Response

We appreciate Reviewer 1's positive summary and detailed methodological comments. These comments were particularly helpful for improving PRISMA transparency, terminology, figure presentation, and the strength of the manuscript's conclusions.

Comment 1

Reviewer comment: The study includes only a PubMed search; at least two databases should be searched.

Response: We agree that multi-database searches are generally preferable for systematic reviews. Because the present review protocol and completed retrieval were restricted to PubMed plus reference-list screening, we did not retrospectively present the search as a multi-database strategy. Instead, we revised the manuscript to report this restriction transparently and to acknowledge the resulting risk of incomplete retrieval and database-selection bias.

Changes made: The Methods now specify PubMed from inception to 12 March 2026 plus reference-list screening, and the Discussion explicitly states that Embase, Web of Science, Scopus, Cochrane Library, and other databases were not searched. We also added PRISMA 2020 and PRISMA-S methodological references.

Comment 2

Reviewer comment: The phrase MCP-1/CCL2 expression does not fully match the content because included datasets used RNA levels, ELISA, optical density, and other outcome measures. MCP-1/CCL2-related readouts may be more appropriate.

Response: We agree. The included outcomes were not limited to a single expression modality, and the previous wording could imply a narrower endpoint than was actually synthesized.

Changes made: The title, Abstract, Methods, Results, Discussion, and Conclusions were revised to use MCP-1/CCL2-related readouts or MCP-1-related readouts where appropriate. The Data Extraction section now clarifies that the synthesized endpoint was an MCP-1/CCL2-related readout rather than one uniform expression measure.

Comment 3

Reviewer comment: Because data are predominantly from rodent models, dose meta-regression is non-significant, heterogeneity is high, and the piglet effect is opposite in direction, strong claims should be avoided.

Response: We agree and have moderated the language throughout the manuscript. The revised text now emphasizes a directionally recurrent but context-dependent preclinical association, rather than a uniform pharmacological effect.

Changes made: Statements such as robustly attenuates, potent and reliable, and heavily mitigating were replaced with more cautious language, including broadly associated, context-dependent, preclinical, and hypothesis-generating. The Abstract and Conclusions now explicitly mention high heterogeneity, single-database retrieval, non-rodent single-dataset estimates, and pharmacokinetic limitations.

Comment 4

Reviewer comment: Figure 1 does not provide reasons for excluding 58 records during title/abstract screening, and the criteria used by the automation tool that excluded 12 studies are unclear.

Response: We agree that the original flow diagram did not provide sufficient screening transparency. We revised both the PRISMA diagram and the accompanying Methods/Results text to clarify the title/abstract exclusion category and the meaning of the automation category.

Changes made: Figure 1 now states that 58 records were excluded because they did not meet title/abstract eligibility criteria related to animal model, resveratrol exposure, or MCP-1/CCL2 outcome. The Methods now explain that automation refers to pre-specified record-level filters applied before manual screening, not exclusion by an opaque machine-learning classifier.

Comment 5

Reviewer comment: Figure 2 could be moved to the Supplementary section.

Response: We agree. The bibliometric visualization was useful for describing the included literature but was not central to the quantitative meta-analysis.

Changes made: The original Figure 2 was moved to Supplementary Figure S1. The Results now contain only a concise descriptive statement, and the main figures were renumbered accordingly.

Comment 6

Reviewer comment: In Figure 3, the term pooled is used for piglets and rabbits, although there is only one dataset for each.

Response: We agree. A single-dataset category should not be described as a pooled subgroup estimate.

Changes made: The forest plot labels were revised from piglet pooled and rabbit pooled to piglet estimate and rabbit estimate. The caption now states that rat and mouse categories are pooled subgroup estimates, whereas piglet and rabbit each represent one dataset.

Comment 7

Reviewer comment: Supplementary Table S5 provides data for only 9 studies, whereas Figure 4B and the meta-regression appear to use the full dataset. The conclusion that dose did not explain heterogeneity is difficult to interpret without a complete dose table.

Response: We agree that the dose analysis requires transparent documentation and should be interpreted cautiously. In the revised manuscript, we avoided presenting the non-significant dose meta-regression as evidence of pharmacological consistency. Instead, we describe it as an exploratory analysis indicating that dose, as modeled in the available dataset, did not explain between-study heterogeneity.

Changes made: The Methods and Results now state that dose was modeled on a log10 scale, that the fitted slope was shallow and non-significant, and that other contributors such as species, disease context, tissue compartment, assay modality, formulation, route of administration, and study quality may dominate the observed variation. The revised supplementary material should provide the complete extracted dose values used for the model in Supplementary Worksheet S5.

Comment 8

Reviewer comment: The heterogeneity data are inconsistent: I2 is reported as 78.9% in the Abstract and Results but as I2 > 80% in the Discussion.

Response: We agree and have corrected the inconsistency.

Changes made: The manuscript now reports the heterogeneity consistently as I2 = 78.9% in the Abstract, Results, figure captions, Discussion, and Conclusions.

Comment 9

Reviewer comment: References [40] and [47] are duplicates of each other; both refer to Cheng et al., BioMed Research International, 2020.

Response: We thank the reviewer for identifying this duplication.

Changes made: The duplicate reference was removed, and subsequent reference numbering and in-text citations were updated. The added methodological references are now numbered [57-63].

Reviewer 2 Report

Comments and Suggestions for Authors

1. The manuscript addresses an important and biologically relevant topic by systematically evaluating the effects of resveratrol on MCP-1/CCL2-related inflammatory signaling in preclinical animal models. The overall concept is interesting and potentially valuable for inflammation-related translational research.
2. The study follows a generally appropriate systematic review and meta-analysis framework, including PRISMA reporting and PROSPERO registration. However, several methodological and interpretative concerns should be addressed before the conclusions can be considered sufficiently robust.
3. The literature search strategy appears limited because only PubMed was searched. Important databases such as Embase, Web of Science, Scopus, or Cochrane Library were not included. This restriction increases the risk of incomplete literature retrieval and potential selection bias.
4. The manuscript repeatedly describes the pooled findings as "robust" and "reliable" despite very high heterogeneity (I² = 78.9%). Such substantial heterogeneity considerably weakens confidence in the pooled effect estimate and should be interpreted with greater caution throughout the manuscript.
5. The pooled standardized mean difference was extremely large (Hedges' g = ‚ąí3.74) . Effect sizes of this magnitude are uncommon in biological systems and may reflect methodological variability, selective reporting, small-study effects, or inconsistent normalization approaches across studies. This issue deserves deeper discussion.
6. The authors acknowledge funnel plot asymmetry but do not perform formal statistical tests for publication bias, such as Egger's regression or Begg's test. Reliance solely on visual inspection is insufficient, especially given the presence of several highly influential datasets.
7. The included studies represent highly heterogeneous disease models, tissues, assays, administration routes, formulations, and dosing regimens. Pooling such biologically diverse conditions into a single summary estimate may reduce mechanistic interpretability and translational relevance.
8. Several analyzable datasets were apparently extracted from the same study when multiple tissues or assays were reported. This approach may violate statistical independence assumptions because datasets originating from the same experiment are likely correlated. The authors should clarify how within-study dependency was handled.
9. The subgroup analyses are underpowered for several categories. For example, the piglet and rabbit subgroups each contain only one dataset, making subgroup-level conclusions unreliable.
10. The dose meta-regression did not identify a significant relationship between resveratrol dose and effect size. This finding weakens claims regarding pharmacological consistency and suggests that other uncontrolled variables may dominate the observed effects.
11. The manuscript strongly emphasizes mechanistic pathways such as AMPK/SIRT1 and NF-kB signalling ; however, these pathways were not systematically analyzed in the included datasets. Consequently, some mechanistic statements appear speculative and extend beyond the direct evidence synthesized in this meta-analysis.
12. The bibliometric analyses and keyword visualizations (Figure 2) contribute little to the study’s scientific conclusions. These sections substantially increase the manuscript's length without providing significant analytical value. Consider condensing or moving some elements to supplementary materials.
13. Risk-of-bias assessment revealed frequent unclear judgments across the allocation concealment, randomization, and blinding domains. These methodological weaknesses should be more explicitly integrated into the interpretation of the pooled findings.
14. No formal certainty-of-evidence framework was applied. Even for preclinical meta-analyses, structured assessment frameworks or adapted GRADE-like approaches could strengthen the interpretation of translational confidence.
15. The manuscript occasionally overstates translational implications. Statements suggesting therapeutic applicability in cardiometabolic disorders should be moderated because the evidence remains entirely preclinical and highly heterogeneous.
16. The discussion appropriately acknowledges the major limitation of poor oral bioavailability of resveratrol. However, this issue substantially limits direct clinical translation and should receive even greater emphasis in the Conclusions.
17. The inclusion of studies using advanced formulations, such as lipid-core nanocapsules or scaffold-based delivery systems, may confound interpretation, as observed effects may not reflect the native pharmacology of resveratrol alone. Stratified analyses according to formulation type would strengthen the manuscript.
18. The cumulative meta-analysis and leave-one-out analyses are useful additions; however, they do not resolve the fundamental issue of biological heterogeneity across the included studies.
19. The Conclusions are stronger than justified by the available evidence. The study demonstrates an overall association between resveratrol exposure and reduced MCP-1-related signaling in animal models, but there is no definitive evidence of reproducible therapeutic efficacy.

Author Response

We thank Reviewer 2 for recognizing the biological relevance of the topic and for providing detailed comments on the methodological and interpretative limitations. We have revised the manuscript to make the conclusions more cautious and to integrate the major sources of uncertainty more explicitly into the interpretation.

Comment 1

Reviewer comment: The literature search strategy is limited because only PubMed was searched; important databases such as Embase, Web of Science, Scopus, or Cochrane Library were not included.

Response: We agree. Rather than overstating the search breadth, we now report the PubMed-only design transparently and present it as a limitation.

Changes made: The Literature Search section now states that PubMed was searched from inception to 12 March 2026 with reference-list screening. The Discussion explicitly acknowledges that studies indexed only in Embase, Web of Science, Scopus, Cochrane Library, or other databases may have been missed.

Comment 2

Reviewer comment: The manuscript repeatedly describes the findings as robust and reliable despite high heterogeneity, which weakens confidence in the pooled estimate.

Response: We agree. The revised manuscript no longer uses robustness as a broad interpretative claim. We distinguish directional stability in sensitivity analyses from biological and methodological robustness.

Changes made: The Abstract, Discussion, and Conclusions were revised to describe the finding as a broad but heterogeneous association. The leave-one-out and cumulative analyses are now interpreted as supporting directional stability, not as resolving heterogeneity.

Comment 3

Reviewer comment: The pooled standardized mean difference is extremely large and may reflect methodological variability, selective reporting, small-study effects, or inconsistent normalization.

Response: We agree that the magnitude of the pooled standardized mean difference requires caution. We expanded the Discussion to address this point directly.

Changes made: The Discussion now notes that the very large pooled standardized mean difference may partly reflect small sample sizes, assay-normalization differences, selective reporting, or influential datasets rather than a single biological effect.

Comment 4

Reviewer comment: The authors acknowledge funnel plot asymmetry but do not perform formal publication-bias tests, such as Egger's regression or Begg's test.

Response: We agree that visual inspection alone is limited. Because the available figure set and extracted manuscript materials did not allow us to implement a reproducible formal asymmetry test within this revision workflow, we did not add an unsupported statistical test. Instead, we expanded the limitation and cited methodological guidance showing that funnel-plot asymmetry can arise from publication bias, heterogeneity, small-study effects, assay differences, or dependency among effect sizes.

Changes made: The Methods and Discussion now state that inference regarding publication bias was limited to qualitative visual inspection. New methodological references on funnel-plot asymmetry and standardized mean differences were added.

Comment 5

Reviewer comment: Pooling highly heterogeneous disease models, tissues, assays, routes, formulations, and dosing regimens into a single summary estimate may reduce mechanistic interpretability and translational relevance.

Response: We agree. The revised manuscript now places greater emphasis on the context-dependent nature of the pooled association and avoids presenting the summary estimate as a universal biological effect.

Changes made: The Discussion now explicitly lists species, disease model, tissue compartment, assay modality, formulation, route of administration, and study quality as contributors to heterogeneity. The Conclusions now state that the evidence remains heterogeneous, preclinical, and methodologically variable.

Comment 6

Reviewer comment: Several analyzable datasets were extracted from the same study, which may violate statistical independence assumptions because datasets from the same experiment are likely correlated.

Response: We agree. The revised Methods now clarify how multiple tissues, assays, or strata were handled and acknowledge that this preserves biological information but may introduce within-study dependency.

Changes made: The Data Extraction section now states that multiple tissues, assay modalities, or experimental strata from a single study were treated as separate analyzable datasets and that such datasets are likely correlated. The Discussion also notes that multilevel or robust-variance approaches may be preferable when sufficient study clusters and covariance information are available.

Comment 7

Reviewer comment: Subgroup analyses are underpowered for several categories; piglet and rabbit each contain only one dataset.

Response: We agree. These categories are no longer described as pooled subgroups.

Changes made: The Results and Figure 2 caption now state that piglet and rabbit represent single-dataset species estimates and should not be interpreted as pooled subgroup effects.

Comment 8

Reviewer comment: The non-significant dose meta-regression weakens claims regarding pharmacological consistency and suggests that uncontrolled variables may dominate the observed effects.

Response: We agree. The revised manuscript now interprets the non-significant dose meta-regression as evidence that dose did not explain between-study heterogeneity in the available dataset, not as evidence of pharmacological consistency.

Changes made: The Results state that the slope was shallow and non-significant. The Discussion and Conclusions now emphasize that uncontrolled variables and pharmacokinetic variability likely contribute to the observed heterogeneity.

Comment 9

Reviewer comment: The manuscript strongly emphasizes AMPK/SIRT1 and NF-kappaB signaling, but these pathways were not systematically analyzed in the included datasets.

Response: We agree that mechanistic statements should be distinguished from the direct meta-analytic endpoint. The revised manuscript now frames these pathways as biologically plausible and literature-consistent, rather than directly demonstrated by the pooled MCP-1/CCL2 dataset.

Changes made: The Discussion now states that AMPK/SIRT1, NF-kappaB, oxidative-stress, and inflammasome-related pathways were not systematically extracted or meta-analyzed in this review.

Comment 10

Reviewer comment: The bibliometric analyses and keyword visualizations contribute little to the scientific conclusions and should be condensed or moved to supplementary materials.

Response: We agree.

Changes made: The bibliometric figure was moved from the main text to Supplementary Figure S1, and the main Results section was condensed accordingly.

Comment 11

Reviewer comment: Risk-of-bias assessment revealed frequent unclear judgments across allocation concealment, randomization, and blinding domains; these weaknesses should be integrated into interpretation.

Response: We agree. The revised manuscript now connects risk-of-bias findings more explicitly to the interpretation of pooled results.

Changes made: The Methods identify the nine assessed risk-of-bias domains, the Results summarize the mixed methodological profile, and the Discussion states that study quality may contribute to between-study heterogeneity and limits certainty.

Comment 12

Reviewer comment: No formal certainty-of-evidence framework was applied; adapted GRADE-like approaches could strengthen translational interpretation.

Response: We agree. Because a formal adapted certainty-of-evidence assessment was not performed in the original analysis, we did not introduce a retrospective certainty grade. Instead, we added a dedicated section explaining this limitation and cited preclinical certainty guidance.

Changes made: A new Certainty of Evidence subsection states that no formal certainty-of-evidence grading was performed and that the evidence should be interpreted as preclinical and hypothesis-generating. A preclinical certainty reference was added.

Comment 13

Reviewer comment: The manuscript occasionally overstates translational implications; therapeutic applicability in cardiometabolic disorders should be moderated.

Response: We agree. The revised manuscript now separates preclinical biological association from clinical therapeutic efficacy.

Changes made: The Discussion and Conclusions now state that the evidence provides a preclinical rationale for further investigation but does not establish reproducible therapeutic efficacy in humans.

Comment 14

Reviewer comment: Poor oral bioavailability of resveratrol substantially limits direct clinical translation and should receive greater emphasis in the Conclusions.

Response: We agree. The Abstract and Conclusions now explicitly mention poor oral bioavailability and pharmacokinetic variability as major limitations that should be addressed before clinical application is inferred.

Comment 15

Reviewer comment: Advanced formulations such as nanocapsules or scaffold-based delivery systems may confound interpretation; formulation-stratified analyses would strengthen the manuscript.

Response: We agree that formulation differences may influence biological effects. However, given the limited number of datasets and the heterogeneity of formulation types, a reliable formulation-stratified analysis was not feasible in the current dataset.

Changes made: The revised Discussion and Conclusions identify formulation and route of administration as sources of heterogeneity and recommend standardized reporting and future formulation-aware analyses.

Comment 16

Reviewer comment: Cumulative meta-analysis and leave-one-out analyses are useful but do not resolve biological heterogeneity.

Response: We agree. The Results and Discussion now state that sensitivity and cumulative analyses support directional stability but do not eliminate the substantial biological and methodological heterogeneity.

Comment 17

Reviewer comment: The Conclusions are stronger than justified; the study shows an association but not definitive reproducible therapeutic efficacy.

Response: We agree and have substantially revised the Conclusions. The Conclusions now state that resveratrol exposure was associated with lower MCP-1/CCL2-related inflammatory readouts in preclinical animal models, but that the evidence remains heterogeneous, preclinical, and methodologically variable and does not establish resveratrol as a reliable therapeutic inhibitor of the MCP-1/CCL2-CCR2 axis in humans. In addition, the revised manuscript now cites methodological sources on PRISMA 2020 explanation and elaboration (Page et al.), PRISMA-S reporting of literature searches (Rethlefsen et al.), interpretation of funnel-plot asymmetry (Sterne et al.), funnel asymmetry testing for standardized mean differences (Pustejovsky and Rodgers), animal-model meta-analysis methods (Yang et al.), robust variance estimation for dependent effect sizes (Hedges, Tipton, and Johnson), and certainty assessment for preclinical animal studies (Hooijmans et al.).

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have nicely addressed all the suggestions made in the previous round. The current form warrants publication.

Reviewer 2 Report

Comments and Suggestions for Authors

Next time, please make all revisions using the track changes mode. It makes the manuscript much easier to review.