Review Reports - Blood–CSF MiRNA Signatures in Alzheimer’s Disease Reveal Novel Biomarker Candidates: A Systematic Review

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This systematic review quantitatively evaluates differential miRNA expression in blood and CSF from patients with AD, provides evidence for previously reported miRNA alterations and also identifies additional candidate miRNAs. However, several methodological issues require further clarification.
1. Study selection: The manuscript mentioned targeted studies in the validation analysis, however, studies focusing exclusively on one or two miRNAs were excluded during the selection process. The rationale for this exclusion is unclear, as it may remove well-designed targeted validation studies that provide important independent evidence for specific miRNAs. The authors should justify this criterion and discuss whether it could bias the assessment of reproducibility.
2. Consistency assessment: The current consistency score (CS), defined as the absolute difference between the numbers of studies reporting opposite directions of change, can be influenced by the total number of studies and may not directly reflect the degree of directional agreement. The purpose of calculating a total CS by combining blood and CSF scores is unclear, particularly when the direction of change is opposite between blood and CSF samples.
3. Biological source: The manuscript broadly groups miRNAs from serum, plasma, whole blood, and other blood fractions under the category of “blood.” However, miRNA profiles can differ across these sample types. The authors should consider analyzing serum, plasma, and other blood-derived samples separately, or at least discuss the potential impact of this heterogeneity on the observed consistency.
4. Strength of evidence assessment: The SES and IES require further methodological justification. Targeted validation studies may not necessarily be methodologically stronger than well-designed discovery studies. Multiplying the CS by the mean SES may therefore introduce artificial precision, particularly if studies reporting the opposite direction of change are not incorporated into the SES calculation.

Author Response

Reviewer 1

We thank the reviewer for the careful evaluation of our systematic review and for recognizing its contribution to the identification of reproducible and novel miRNA candidates in Alzheimer’s disease. We appreciate the methodological concerns raised, which have prompted us to further clarify several aspects of the study design, evidence synthesis, and interpretation of the results.

Study selection: The manuscript mentioned targeted studies in the validation analysis, however, studies focusing exclusively on one or two miRNAs were excluded during the selection process. The rationale for this exclusion is unclear, as it may remove well-designed targeted validation studies that provide important independent evidence for specific miRNAs. The authors should justify this criterion and discuss whether it could bias the assessment of reproducibility:

We agree that the rationale for excluding studies focused exclusively on one or two miRNAs was not sufficiently explained in the original manuscript.

The primary objective of this systematic review was to identify miRNAs emerging from biomarker-discovery or multi-candidate profiling approaches, rather than to evaluate the accumulated evidence for individual miRNAs selected a priori. Studies assessing only one or two miRNAs generally investigate candidates already implicated in AD or selected on the basis of previous biological or clinical evidence, investigator-driven hypotheses, or laboratory-specific research priorities. Including these studies within a discovery-oriented framework could therefore introduce candidate-selection bias and potentially increase the apparent reproducibility of miRNAs that have already been repeatedly selected for targeted investigation.

Importantly, targeted approaches were not excluded per se. Studies evaluating three or more miRNAs were eligible, including targeted multi-miRNA studies, provided that they fulfilled the remaining inclusion criteria. We have revised the Eligibility Criteria section to make this distinction and the rationale for the ≥3-miRNA threshold explicit.

Although, as pointed before, including such studies could introduce candidate-selection bias, we also agree that excluding them may omit well-designed independent validation studies providing relevant evidence for specific miRNAs. We have therefore acknowledged this point as a limitation and clarified that the reproducibility estimates obtained in this review reflect the biomarker-discovery and multi-candidate profiling literature, rather than an exhaustive assessment of all available targeted validation evidence for each individual miRNA. We have also made several corrections on the results and discussion to clarify this point.

Consistency assessment: The current consistency score (CS), defined as the absolute difference between the numbers of studies reporting opposite directions of change, can be influenced by the total number of studies and may not directly reflect the degree of directional agreement.

The purpose of calculating a total CS by combining blood and CSF scores is unclear, particularly when the direction of change is opposite between blood and CSF samples.

We thank the reviewer for this important observation. We agree that the CS should not be interpreted as a normalized measure of directional agreement, because the total number of studies reporting each miRNA is not used as a denominator. Thus, identical CS values may arise from different proportions of studies supporting the predominant direction. However, normalization by the total number of studies would shift the emphasis toward proportional agreement and could assign maximal values to miRNAs supported by only a small number of studies. For example, miR-454-3p, reported in three studies with fully concordant directions, would obtain a normalized value of 1, whereas miR-532-5p, reported in six studies with only one study opposing the predominant trend, would obtain a lower normalized value of 0.67. Thus, normalization alone would not capture the extent of independent replication.

Rather, the CS was designed as a descriptive measure of the net directional support across independent studies, calculated as the difference between the numbers of studies reporting the predominant and opposite trends. Consequently, the score incorporates the absolute numbers of supporting and opposing studies but does not represent a normalized agreement proportion. We have revised the Methods section to make this interpretation explicit.

We have also clarified the rationale for combining blood and CSF scores. The cumulative CS is not intended to measure directional concordance between the two biofluids. Rather, it summarizes the net directional support independently within blood and within CSF. This distinction is particularly relevant to the rationale of our cross-biofluid analysis, since miRNAs altered in CSF may also be detectable and dysregulated in blood, a more accessible and less invasive biofluid, without necessarily showing the same direction of change in both compartments. Such differences may reflect compartment-specific regulatory mechanisms and biofluid dynamics. Therefore, a miRNA may show a high cumulative CS even when its predominant direction is opposite between blood and CSF, provided that the predominant direction within each biofluid is supported by multiple independent studies. Concordant versus discordant blood-CSF expression patterns were evaluated separately and are explicitly reported in the manuscript.

To further improve transparency, Supplementary Figure 1 has also been revised to provide a detailed schematic representation of how both the SES and CS are calculated and interpreted.

Biological source: The manuscript broadly groups miRNAs from serum, plasma, whole blood, and other blood fractions under the category of “blood.” However, miRNA profiles can differ across these sample types. The authors should consider analyzing serum, plasma, and other blood-derived samples separately, or at least discuss the potential impact of this heterogeneity on the observed consistency.

We agree that serum, plasma, whole blood, and other blood-derived fractions may show different miRNA profiles. In our primary analysis, these sample types were grouped under the broader category of “blood” to enable the cross-biofluid comparison with CSF. However, the specific sample type used in each study was recorded and is reported in Supplementary Table 1A.

We have now clarified this grouping in the Methods section (Cross-study integration and consistency assessment section) and explicitly acknowledged in the Discussion that heterogeneity among blood-derived sample types may contribute to variability in miRNA profiles and influence the observed consistency across studies (last paragraph of the discussion).

Strength of evidence assessment: The SES and IES require further methodological justification. Targeted validation studies may not necessarily be methodologically stronger than well-designed discovery studies. Multiplying the CS by the mean SES may therefore introduce artificial precision, particularly if studies reporting the opposite direction of change are not incorporated into the SES calculation.

We thank the reviewer for this important methodological observation. We agree that targeted validation studies should not necessarily be interpreted as methodologically stronger than well-designed discovery studies. We have therefore revised the interpretation and terminology of the SES. The SES is now described as a Study Evidence Score and is used descriptively to characterize study design and validation support, rather than as a formal measure of methodological quality or certainty of evidence.

We also agree that mathematically combining the mean SES with the CS may introduce an unwarranted level of numerical precision. We have therefore completely removed the Integrated Evidence Score (IES) and the corresponding quantitative ranking from the manuscript.

In the revised analysis, net directional support and study design/validation support are treated as two complementary dimensions. The CS incorporates studies reporting both the predominant and opposite directions of expression and therefore captures directional inconsistency. In contrast, the mean SES is calculated from studies supporting the predominant expression pattern and is used descriptively to characterize their study design and validation support. These two measures are now displayed separately in the revised Figure 3 and are no longer mathematically combined.

Supplementary Figure 1 has been revised to separately illustrate the SES classification and mean SES calculation (Panel A) and the CS calculation, including predominant and opposite directions of change (Panel B). The former IES calculation has been removed, emphasizing that SES and CS represent complementary but distinct dimensions of the evidence and are interpreted separately.

Reviewer 2 Report

Comments and Suggestions for Authors

Overall :
The manuscript addresses a relevant and potentially valuable topic by systematically reviewing the role of microRNAs in Alzheimer’s disease using both blood- and cerebrospinal fluid (CSF)-based evidence. The manuscript is generally well structured, and the review appears to follow the PRISMA 2020 framework. However, several aspects could be strengthened to better establish the clinical relevance of microRNAs, the methodological rigor of the systematic review, and their potential added value over established Alzheimer’s disease biomarkers such as amyloid-β (Aβ) and tau.

1. Abstract

The abstract is generally well presented, with a clear description of the study focus, major findings, and conclusion. However, the conclusion could be strengthened by briefly indicating the potential clinical significance and current limitations of microRNA as a biomarker, particularly whether the evidence supports its use primarily for early detection, disease monitoring, diagnosis, or prognosis.

2. Introduction

The introduction provides a good rationale for examining microRNAs in both blood and CSF. However, the specific importance of microRNAs as biomarkers for Alzheimer’s disease should be more clearly established.

In particular, the authors should clarify:

Why are microRNAs considered promising biomarkers for Alzheimer’s disease?
Is their principal potential application early detection, diagnosis, disease monitoring, prognosis, or prediction of disease progression?
What biological characteristics of microRNAs make them potentially advantageous as biomarkers, such as their stability in biological fluids, regulatory functions, or relationship with disease-related pathways?
More importantly, how might microRNA-based biomarkers complement or improve upon currently established biomarkers such as Aβ42/Aβ40, phosphorylated tau (p-tau), total tau, and neurofilament light chain (NfL)?
The rationale for reviewing both blood and CSF should also be further developed, particularly regarding the potential differences between these two sample types in terms of accessibility, disease specificity, and clinical applicability.

A clearer explanation of these points would help establish the knowledge gap and clinical rationale for conducting the systematic review.

3. Methods

The Methods section generally follows the principles of systematic review methodology. Nevertheless, the methodological framework could be reported more explicitly to improve reproducibility.

The authors should clearly define the PICO/PICOS elements, particularly:

Population: diagnostic groups and relevant disease stages, e.g., cognitively normal individuals, subjective cognitive decline, mild cognitive impairment (MCI), and Alzheimer’s disease.
Comparator: healthy controls, non-Alzheimer’s dementia, MCI, or other relevant neurological conditions, where applicable.
Outcomes: specific microRNAs, diagnostic accuracy, differential expression, sensitivity/specificity, association with disease severity, progression, or other clinical outcomes.

In addition, the authors should clearly specify which databases were searched, or any restrictions applied, such as language or publication date. If applicable, the complete search strategy for at least one database should be provided as supplementary material.

It would also be useful to clarify whether the study selection and data extraction were conducted independently by two reviewers and how disagreements were resolved.

4. Results

The Results section is generally well organized and appears to consistently follow the PRISMA 2020 reporting framework, particularly in presenting the study-selection process.

However, the authors should consider whether heterogeneity among the included studies should be more explicitly evaluated and reported. Given the likely differences in:

participant characteristics and disease stages,
blood versus CSF sample types,
sample collection and processing methods,
microRNA detection platforms,
normalization/reference methods,
candidate versus high-throughput microRNA profiling,
and diagnostic criteria,

substantial methodological and clinical heterogeneity may exist.

The authors should therefore clarify whether heterogeneity was assessed and, if applicable, discuss how it influences the interpretation of the findings. If quantitative meta-analysis was not appropriate because of substantial heterogeneity, this should be explicitly justified.

5. Discussion

The Discussion would benefit from a stronger emphasis on the clinical implications of the findings.

early biomarkers before significant cognitive decline,
diagnostic biomarkers for distinguishing Alzheimer’s disease from controls or other dementias,
monitoring biomarkers for disease progression or treatment response, or
prognostic biomarkers for predicting clinical progression.

Importantly, the manuscript should place the findings in the context of established Alzheimer’s disease biomarkers, including CSF Aβ and tau and emerging blood-based biomarkers such as plasma p-tau and NfL. The authors should discuss whether microRNAs provide potential additional information, rather than simply being another biomarker candidate.

The discussion could particularly address the potential advantages and limitations of blood-based microRNAs compared with CSF biomarkers. Blood sampling is substantially less invasive and potentially more suitable for large-scale screening, but peripheral microRNA profiles may be influenced by systemic conditions and other biological factors. Therefore, the authors should critically discuss whether the current evidence is sufficient to support clinical translation.

Author Response

The manuscript addresses a relevant and potentially valuable topic by systematically reviewing the role of microRNAs in Alzheimer’s disease using both blood- and cerebrospinal fluid (CSF)-based evidence. The manuscript is generally well structured, and the review appears to follow the PRISMA 2020 framework. However, several aspects could be strengthened to better establish the clinical relevance of microRNAs, the methodological rigor of the systematic review, and their potential added value over established Alzheimer’s disease biomarkers such as amyloid-β (Aβ) and tau.

We thank the reviewer for the positive assessment of the relevance and overall structure of the manuscript. We also appreciate the constructive suggestions regarding the clinical relevance of miRNAs, the methodological reporting of the systematic review, and their relationship with established Alzheimer’s disease biomarkers. These comments have helped us strengthen the Introduction, Methods, and Discussion and to better define the potential clinical role and current limitations of miRNA-based biomarkers. We believe that all the reviewer’s comments have been carefully addressed in the revised manuscript.

Abstract

We thank the reviewer for this helpful suggestion. We have revised the conclusion of the Abstract to clarify both the potential clinical relevance and the current limitations of miRNAs as AD biomarkers. Because the studies included in our review primarily compared patients with AD with healthy controls, the available evidence most directly supports further evaluation of miRNAs as diagnostic/discriminative biomarkers. In contrast, the current evidence is insufficient to establish their utility for early detection, longitudinal disease monitoring, or prognosis. We have also emphasized that clinical translation remains limited by biofluid and methodological heterogeneity and requires harmonized cross-cohort and longitudinal validation.

Introduction

The introduction provides a good rationale for examining microRNAs in both blood and CSF. However, the specific importance of microRNAs as biomarkers for Alzheimer’s disease should be more clearly established.

We have expanded the Introduction to more clearly establish the biological and potential clinical relevance of miRNAs as biomarkers in Alzheimer’s disease. Specifically, the revised text now addresses the following points:

In particular, the authors should clarify:

Why are microRNAs considered promising biomarkers for Alzheimer’s disease?

We now explain that miRNAs can be detected in both blood and CSF and may provide information on epigenetic alterations associated with AD. Their potential accessibility in peripheral biofluids, together with their relationship with disease-associated molecular pathways, makes them attractive candidates for biomarker development.

Is their principal potential application early detection, diagnosis, disease monitoring, prognosis, or prediction of disease progression?

We have clarified that, because most of the studies included in this systematic review compare patients with AD with cognitively healthy controls, the current evidence primarily supports the further evaluation of miRNAs as diagnostic/discriminative biomarkers. Their potential utility for preclinical or early detection, longitudinal disease monitoring, prognosis, or prediction of disease progression remains less established and will require specifically designed longitudinal studies.

What biological characteristics of microRNAs make them potentially advantageous as biomarkers, such as their stability in biological fluids, regulatory functions, or relationship with disease-related pathways?

The Introduction now highlights the relative stability of miRNAs in biological fluids, particularly when associated with extracellular vesicles or RNA-binding complexes, as well as their regulatory role in pathways relevant to AD, including amyloid processing, tau-related mechanisms, neuroinflammation, synaptic dysfunction, and neuronal survival.

More importantly, how might microRNA-based biomarkers complement or improve upon currently established biomarkers such as Aβ42/Aβ40, phosphorylated tau (p-tau), total tau, and neurofilament light chain (NfL)?

We have also clarified that miRNA-based biomarkers should currently be regarded as complementary rather than alternative to established biomarkers such as Aβ42/Aβ40, phosphorylated tau, total tau, and NfL. Whereas these established biomarkers provide information on core AD pathology and neurodegeneration, miRNAs may provide an additional regulatory and pathway-related molecular layer that could contribute to future multimarker approaches. However, their added clinical value still requires validation using standardized methods, independent cohorts, and longitudinal studies.

These additions have been incorporated into the Introduction to better define both the potential value and the current limitations of miRNAs as AD biomarkers.

The rationale for reviewing both blood and CSF should also be further developed, particularly regarding the potential differences between these two sample types in terms of accessibility, disease specificity, and clinical applicability. A clearer explanation of these points would help establish the knowledge gap and clinical rationale for conducting the systematic review.

We have revised the Introduction to better explain the complementary rationale for analyzing blood and CSF. Blood offers greater accessibility, minimal invasiveness, and suitability for repeated sampling, whereas CSF, although more invasive, may more directly reflect CNS-related molecular changes. We also now emphasize that it remains unclear which AD-associated miRNA alterations are shared between both biofluids and which are compartment-specific, providing the clinical and biological rationale for our cross-biofluid analysis.

Methods

The Methods section generally follows the principles of systematic review methodology. Nevertheless, the methodological framework could be reported more explicitly to improve reproducibility.

The authors should clearly define the PICO/PICOS elements, particularly:

Population: diagnostic groups and relevant disease stages, e.g., cognitively normal individuals, subjective cognitive decline, mild cognitive impairment (MCI), and Alzheimer’s disease. Comparator: healthy controls, non-Alzheimer’s dementia, MCI, or other relevant neurological conditions, where applicable. Outcomes: specific microRNAs, diagnostic accuracy, differential expression, sensitivity/specificity, association with disease severity, progression, or other clinical outcomes.

We have revised the Methods section to explicitly define the review question according to a PICOS framework. The population comprised individuals with Alzheimer’s disease; the index of interest was miRNA expression in blood-derived samples or CSF; the comparator consisted of age-matched cognitively healthy controls; the primary outcomes were differential miRNA expression, direction of dysregulation, biofluid distribution, and reproducibility across independent studies; and eligible study designs included human observational studies using discovery or multi-candidate miRNA profiling approaches.

The Methods section already specifies that PubMed was used as the information source, that the search was conducted initially in February 2025 and updated in March 2026, and provides the complete search strategy used. The eligibility criteria also specify that studies published within the previous 15 years were considered. In addition, the use of a single database is acknowledged in the Discussion as a limitation of the review. We have now further clarified whether any language restrictions were applied.

It would also be useful to clarify whether the study selection and data extraction were conducted independently by two reviewers and how disagreements were resolved.

The study selection procedure was already described as being conducted independently by two reviewers, with disagreements resolved through discussion and additional reviewer involvement when necessary. We have now clarified that data extraction was also performed independently by two reviewers, with discrepancies resolved by consensus and consultation of a third member of the review team when required.

Results

The Results section is generally well organized and appears to consistently follow the PRISMA 2020 reporting framework, particularly in presenting the study-selection process.

However, the authors should consider whether heterogeneity among the included studies should be more explicitly evaluated and reported. Given the likely differences in: participant characteristics and disease stages, blood versus CSF sample types, sample collection and processing methods, microRNA detection platforms, normalization/reference methods, candidate versus high-throughput microRNA profiling and diagnostic criteria, substantial methodological and clinical heterogeneity may exist.

We thank the reviewer for this important comment. Heterogeneity among the included studies was assessed descriptively based on the extracted study characteristics, including participant and cohort characteristics, biological source and sample type, miRNA detection technology, study design, diagnostic criteria, APOE assessment, and statistical or computational methods. These variables are reported in Supplementary Table 1A and summarized in the Results. Given the substantial methodological heterogeneity across studies, a conventional quantitative meta-analysis was considered inappropriate; therefore, we used a direction-of-effect and consistency-based synthesis rather than pooled effect estimates. We have also expanded the Discussion to clarify how this heterogeneity may influence the interpretation of the findings and to acknowledge additional variability related to sample collection and processing procedures and miRNA normalization/reference strategies, which were not consistently reported across studies.

Discussion

The Discussion would benefit from a stronger emphasis on the clinical implications of the findings.

At present, the discussion should go beyond identifying differentially expressed microRNAs and consider what these findings could potentially mean for clinical practice. In particular, the authors should discuss whether the reviewed evidence supports microRNAs as: early biomarkers before significant cognitive decline, diagnostic biomarkers for distinguishing Alzheimer’s disease from controls or other dementias, monitoring biomarkers for disease progression or treatment response, or prognostic biomarkers for predicting clinical progression.

We thank the reviewer for this important suggestion. This point has now been addressed not only in the Introduction and Abstract but also more explicitly in the Discussion. We clarify that, because the included evidence predominantly compares patients with AD with cognitively healthy controls, the current evidence primarily supports further evaluation of miRNAs as diagnostic/discriminative biomarkers. Evidence supporting their use for preclinical detection, longitudinal monitoring, treatment response, or prognosis remains limited. We have also placed our findings in the context of established CSF and blood-based AD biomarkers, emphasizing that miRNAs should be considered as potentially complementary markers that may provide an additional regulatory and pathway-related molecular layer, rather than as replacements for Aβ-, tau-, or NfL-based biomarkers. We further emphasize that their added clinical value requires direct multi-marker comparison and longitudinal validation.

We thank the reviewer for this important comment. We have expanded the Discussion to more explicitly contrast the potential clinical advantages and limitations of blood- and CSF-based miRNA biomarkers. Blood-based miRNAs offer clear practical advantages because sampling is minimally invasive and suitable for repeated measurements and large-scale application, although peripheral miRNA profiles may also be influenced by systemic and pre-analytical factors and may not directly reflect CNS-specific changes. In contrast, CSF is more invasive to obtain but may provide molecular information more closely related to CNS processes. We therefore emphasize that, despite their potential, the current evidence is not yet sufficient to support the clinical implementation of miRNA-based biomarkers and that standardized, longitudinal, and multimarker validation studies are still required.

Reviewer 3 Report

Comments and Suggestions for Authors

Your systematic review provides a rigorous, cross-compartmental analysis of miRNA signatures in Alzheimer’s disease. By analyzing 90 studies and differentiating between blood and CSF profiles, you have offered much-needed clarity in a highly heterogeneous field. The conceptualization of the Consistency Score (CS), Strength of Evidence Score (SES), and Integrated Evidence Score (IES) is an elegant workaround to the limitations that typically preclude quantitative meta-analyses of such diverse datasets. The manuscript is well-written, and the findings are synthesized into actionable insights for the biomarker research community. Please address the following comments to refine the manuscript for publication.

Major Comments

Validation of the Integrated Evidence Score (IES): The IES is a practical and creative metric to evaluate the selected miRNAs. However, please briefly clarify in the Discussion whether this scoring approach was adapted from a specific precedent in the literature or if it is entirely novel to this study. If it is novel, explicitly state this and briefly acknowledge any inherent limitations of this ranking tool (e.g., potential weighting biases) in the "Study Limitations" section.
APOE Genotype Discussion: In Section 3.1 (Line 282), you note that 21 studies reported APOE genotype information and indicated that it "did not influence the results". Given that APOE is the strongest genetic risk factor for sporadic AD, this is a surprising and important observation. Expand briefly in the Discussion on why circulating/CSF miRNA levels might be independent of APOE status based on these studies, or if this lack of influence is potentially due to small sample sizes or lack of statistical power in the primary literature.
Software and Database Specifications: In Sections 2.3, 2.9, and 2.10, provide the exact version numbers for the software and databases used at the time of your analysis (e.g., Rayyan AI, R package multiMiR, MirTarBase, RNA-Disease Repository) to ensure full reproducibility.

Minor Comments & Typographical Corrections

Line 45 & 47: Citation format spacing issues. Replace [2.3] with [2,3] and [4.5] with [4,5].
Line 71: Typographical error. The phrase "...commonly used in clinical to study..." should likely read "...commonly used in clinical practice to study..." or "...used in clinics to study...".
Line 347 (Table 2C Header): There is a typographical error in the table sub-header: "Increasedecrease/" should be corrected to "Increase/Decrease".
Table 3 Formatting: Several columns in Table 3 (specifically regarding the CS values and the trend arrows) appear misaligned or floating in the manuscript draft. Please ensure standard cell borders and alignments are applied for readability prior to final typesetting.
Line 635: Missing punctuation. Add a period after the citation [111-113] before the start of the next sentence ("Therefore...").
Figure 1 (PRISMA Flow Diagram): The mathematical flow is correct (247 - 151 - 6 = 90), but ensure high-resolution files are provided for publication, as some of the smaller text in the screening boxes may become blurry upon journal formatting.

Author Response

We sincerely thank the reviewer for the positive assessment of our systematic review and for recognizing the value of the cross-compartmental approach used to integrate miRNA findings from blood and CSF. We particularly appreciate the reviewer’s comments regarding the clarity of the synthesis in such a heterogeneous field. The constructive suggestions provided have helped us further refine the methodological description, interpretation of the findings, and presentation of the results.

Major Comments

Validation of the Integrated Evidence Score (IES): The IES is a practical and creative metric to evaluate the selected miRNAs. However, please briefly clarify in the Discussion whether this scoring approach was adapted from a specific precedent in the literature or if it is entirely novel to this study. If it is novel, explicitly state this and briefly acknowledge any inherent limitations of this ranking tool (e.g., potential weighting biases) in the "Study Limitations" section.

We thank the reviewer for this comment. We have clarified in the revised manuscript that both the Consistency Score (CS) and the Study Evidence Score (SES) are descriptive metrics developed specifically for this systematic review and do not correspond to previously validated standardized scoring systems. To improve transparency, Supplementary Figure 1 has also been revised to explicitly illustrate how both scores are calculated and interpreted. In addition, because the Integrated Evidence Score (IES) was removed from the revised analysis, the original IES-based figure was replaced by a more informative visualization in which CS and mean SES are displayed separately. The new figure additionally incorporates the number of supporting studies and the predominant direction of change in blood and CSF, allowing readers to evaluate reproducibility, study support, and cross-biofluid expression patterns without reducing these dimensions to a single composite score.

APOE Genotype Discussion: In Section 3.1 (Line 282), you note that 21 studies reported APOE genotype information and indicated that it "did not influence the results". Given that APOE is the strongest genetic risk factor for sporadic AD, this is a surprising and important observation. Expand briefly in the Discussion on why circulating/CSF miRNA levels might be independent of APOE status based on these studies, or if this lack of influence is potentially due to small sample sizes or lack of statistical power in the primary literature.

We agree that the original wording could be interpreted too broadly. Although 21 studies reported APOE genotype information, only several specifically indicated that APOE did not significantly influence their miRNA findings, and APOE was not consistently analyzed as an effect modifier across studies. We have therefore clarified this statement in the Results (Search strategy section) and expanded the Discussion (final paragraph of miRNAs with shared dysregulation in blood and CSF in AD section) to emphasize that the absence of significant APOE effects should not be interpreted as evidence of independence from APOE status. Limited statistical power for genotype-stratified analyses and methodological heterogeneity may contribute to these negative findings, while genotype-specific effects have also been reported in the literature included in our review. We therefore highlight the need for larger studies specifically designed to assess APOE-dependent miRNA profiles.

Software and Database Specifications: In Sections 2.3, 2.9, and 2.10, provide the exact version numbers for the software and databases used at the time of your analysis (e.g., Rayyan AI, R package multiMiR, MirTarBase, RNA-Disease Repository) to ensure full reproducibility.

We thank the reviewer for this suggestion. We have revised Sections 2.3, 2.9, and 2.10 to provide the relevant software and database version information. Because Rayyan is a continuously updated web platform and does not provide a fixed version number, we have reported the platform and access date. For the bioinformatic analyses, we have specified the version of the multiMiR R package and its underlying database, the miRTarBase release integrated within multiMiR, and RNADisease v4.0.

Minor Comments & Typographical Corrections

Line 45 & 47: Citation format spacing issues. Replace [2.3] with [2,3] and [4.5] with [4,5].

The citation formatting has been corrected

Line 71: Typographical error. The phrase "...commonly used in clinical to study..." should likely read "...commonly used in clinical practice to study..." or "...used in clinics to study...".

This paragraph has been replaced as a consequence of reviewer’s comments.

Line 347 (Table 2C Header): There is a typographical error in the table sub-header: "Increasedecrease/" should be corrected to "Increase/Decrease".

This error has been corrected.

Table 3 Formatting: Several columns in Table 3 (specifically regarding the CS values and the trend arrows) appear misaligned or floating in the manuscript draft. Please ensure standard cell borders and alignments are applied for readability prior to final typesetting.

This appears to be a proof-related formatting error. We will correct it if it happens again when the newly revised manuscript is submitted.

Line 635: Missing punctuation. Add a period after the citation [111-113] before the start of the next sentence ("Therefore...").

This issue has been corrected.

Figure 1 (PRISMA Flow Diagram): The mathematical flow is correct (247 - 151 - 6 = 90), but ensure high-resolution files are provided for publication, as some of the smaller text in the screening boxes may become blurry upon journal formatting.

We have uploaded a higher-resolution version of the figure.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the authors` responses to my previous comments and the revisions made to the manuscript. All of my concerns have been satisfactorily addressed. I recommend acceptance.