supplementary figure

supplementary materials

eTables

Author response:
We appreciate the constructive and thoughtful reviews provided by the reviewers and editorial team. We thank you for the opportunity to submit a provisional response and are grateful for the detailed and critical feedback that will strengthen our work. Below, we provide a summary of our planned revisions in response to the public reviews from Reviewer #1 and Reviewer #2.
Reviewer #1 – Public Review Response Plan
(1) Sample Overlap (MR Bias):
We plan to replace several non-overlapping GWAS data sources to validate the association between aneurysms and atherosclerosis, thereby eliminating bias and Type I errors caused by sample overlap.
(2) Multivariable MR (MVMR): 
We will attempt to incorporate known confounding factors (e.g., hypertension, smoking, diabetes) within the multivariable MR framework to verify the robustness of our results.
(3) Clarifications and Presentation:
- We will correct eTable citations.
- Distinguish correctly between "incidence" and "prevalence".
- Reorganize results to consistently present primary analyses first (IVW), followed by sensitivity results.
- Expand the methods section to fully reflect all analyses.
Reviewer #2 – Public Review Response Plan
(1) Justification of MMP Selection: 
We will provide a detailed rationale for the inclusion of the 12 MMPs, based on prior literature and biological relevance.
(2) Multiple Testing Clarification: 
We will clarify the Bonferroni correction strategy, explicitly accounting for all tests (e.g., 72 comparisons × multiple MR methods).
(3) Instrument Selection Threshold:
- We agree with the reviewer and will revise the SNP selection strategy, starting from p &lt; 5×10⁻⁸ and only relaxing thresholds when fewer than 3 instruments are found.
- Clarify the reasons why we do not use LD proxies.
(4) Pleiotropy and Heterogeneity Tests:
- We will add Egger's intercept results alongside MR-PRESSO.
- Specify the R packages used (e.g., TwoSampleMR).
- To prevent cluttered data presentation, we have included both heterogeneity and pleiotropy p-values in the supplementary tables.
- Supplement forest plots showing outlier exclusion effects.
(5) Clarifications in Figures and Tables:
- Fix the duplicated “simple mode” entry in Figure 2.
- Correct inconsistencies in p-values between figures and text.
- Improve figure legends (e.g., color bar labels, panel identifiers).
- Revise Table 4 title for clarity.
- Remove the term "causal" where associations are nominal (e.g., p ~ 0.05).

Anurag Agrawal

Balram Bhargava

eLife Assessment
This is valuable work with theoretical implications for possible mediation by MMP12 in the link between atherosclerosis and intracranial aneurysms, using Mendelian Randomization for causal inference. Additional analysis would be required to verify the claims, which currently have incomplete support in terms of the strength of evidence. Given that most of the identified causal associations do not hold after correcting for multiple tests, the conclusions should be carefully reviewed in order to be fully supported by the results.

anonymous

Reviewer #1 (Public review):
Summary:
The authors performed bidirectional two-sample Mendelian randomization using publicly available GWAS summary data to assess the directional causal association between atherosclerosis and intracranial aneurysms. They have used a similar strategy to identify the role of matrix metalloproteinases (MMP), especially MMP12, in mediating the above causal association. They finally substantiated these results by measuring and comparing the MMP12 levels in the plasma samples collected from carotid atherosclerosis and intracranial aneurysm patients with those of healthy controls. Local tissue levels of MMP12 were also measured in experimental mouse models.
Strengths:
The authors have chosen to address an important problem that could be of interest to many researchers and clinicians in the subfield.
Weaknesses:
Mendelian Randomization (MR) is a powerful approach to explore the directional causal relationship between comorbid conditions using genetic variants as instrumental variables. The validity of causal inference derived from MR strongly depends on genetic instruments satisfying the three core assumptions- relevance, independence, and exclusion restriction. The violation of these assumptions is hard to verify in many real-world situations and may result in spurious results. Rigorous sensitivity analysis is essential to ensure the robustness of the results. The sensitivity analysis presented in the current manuscript is incomplete. The key points are as follows:
(1) The GWAS summary datasets used by the authors for assessing the causal relationship between atherosclerosis and intracranial aneurysms were all from the FinnGen study and thus may have overlapping samples which is known to introduce bias into the causal estimates and inflate type 1 error rates.
(2) Both atherosclerosis and aneurysms share common risk factors (mentioned by the authors as well) such as hypertension, cholesterol, diabetes, smoking, etc., which could lead to correlated pleiotropy while performing Mendelian randomization. MR-PRESSO may not effectively account for the same.
(3) The authors explored the role of matrix metalloproteinases as intermediate biomarkers mediating the risk of atherosclerosis in the intracranial aneurysms. Separating the exposure to biomarker MR from biomarker to outcome MR limits the interpretation of the results. The effect size of the indirect effect cannot be assessed.
(4) The scatter plots presented in Supplementary Figures 1-3 are neither cited nor discussed in the manuscript. Some of the plots show variability in the direction and magnitude of the causal estimates from MR-Egger and MR-IVW methods, indicating either masking of the causal estimates or directional pleiotropy. Discussing these results is crucial to inform the readers of the limitations of the derived causal estimates.
(5) When there is substantial evidence available for the frequent coexistence of atherosclerosis and aneurysms, the additional value of the cross-sectional data showing the increased prevalence of atherosclerosis in patients with intracranial aneurysms without adjusting for confounding risk factors is not clear.
(6) It is also not clear from the manuscript whether the authors are projecting the MMP12 as a shared biomarker or as a mediator between atherosclerosis and intracranial aneurysms. As also noted by the authors, assessment of plasma MMP12 levels in a cross-sectional sample is not sufficient to substantiate the role of MMP12 as an intermediate biomarker connecting atherosclerosis to the increased risk of intracranial aneurysms.
Impact:
The findings from this study can form the basis for a more systematic analysis towards identifying molecular intermediates mediating the risk of atherosclerosis in patients with intracranial aneurysms or vice versa, which in turn helps develop novel strategies to manage these comorbid conditions.

Reviewer #2 (Public review):
The manuscript by Liu and colleagues applied Mendelian Randomization (MR) techniques to study the causal relationship of atherosclerosis (categorized into four subtypes) and intracranial aneurysms (classified as unruptured or ruptured), as well as the potential mediation by 12 plasma matrix metalloproteinase (MMP) levels. The authors have followed rigorous MR analysis guidelines by using multiple analytical approaches, implementing strict selection criteria, and employing comprehensive sensitivity analyses. One of the strengths is the lack of overlapping samples in their two-sample MR analysis. This approach helps mitigate potential biases and increases the reliability of their causal inference. The analysis is fundamentally sound, but there are still several nuanced areas where the methodology could be strengthened. Given that most of the identified causal associations do not hold after correcting for multiple tests, the conclusions should be carefully reviewed to be fully supported by the results.
The recommendations below are meant to enhance the already robust approach.
(1) The selection of 12 MMPs lacks a clear, explicit rationale in the provided excerpt. A more detailed explanation of why these specific MMPs were chosen would strengthen the methodological rigor.
(2) Adjusting p-value for multiple testing using Bonferroni correction needs to be elucidated better.
(3) The authors should provide a more robust explanation of why they shifted from 5×10-9 to 5×10-6 to select genomic instruments.
(4) Egger's intercept may be a more robust approach for this study to test horizontal pleiotropy rather than MR-PRESSO.

Atherosclerosis, Intracranial Aneurysms, and Intermediate Biomarkers: Real-World Observational and Mendelian Randomization Research

Figures and data

Research arrangement.

Cross-sectional study baseline table.

Sources of Mendelian randomization genetic instruments.

Two-way MR analysis of the associations between atherosclerosis and intracranial aneurysms (P < 0.05).

Two-way MR analysis of the associations between circulating MMPs levels and atherosclerosis and intracranial aneurysms (P < 0.05).

Baseline table of patients with atherosclerosis and intracranial aneurysm serum.

Logistic regression analysis of the associations between serum MMP12 levels and atherosclerosis and intracranial aneurysms.

Be the first to read new articles from eLife