Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents a two-sample Mendelian randomization study investigating the causal relationships between immune cell phenotypes, plasma metabolites, and pericarditis. The integration of immunophenotypic and metabolomic datasets within a mediation MR framework is innovative and potentially relevant to the understanding of pericarditis pathophysiology. Overall, the study addresses an important clinical question and employs contemporary genetic epidemiology methods. Nevertheless, several methodological, analytical, and interpretative issues currently limit the strength of the conclusions and should be addressed before publication in a high-impact journal.

The title is attractive and concise; however, its wording suggests a degree of mechanistic certainty that exceeds the evidence generated by Mendelian randomization. The statement that a "B Cell-Cholate Pathway Mediates the Genetic Risk of Pericarditis" implies that the pathway has been definitively demonstrated. In reality, the study provides genetic evidence supporting a potential mediation mechanism rather than direct biological proof. A more cautious title emphasizing the identification of a putative immunometabolic axis would be more appropriate and aligned with current MR reporting standards.

The abstract is generally well organized and clearly summarizes the analytical workflow. However, a critical concern relates to the absence of any reference to correction for multiple testing. The study evaluates 731 immune phenotypes and 1,400 metabolites, generating an extremely high-dimensional analytical setting. Yet significance is reported using nominal P-values. Readers are left unable to determine which findings remain statistically robust after controlling for multiple comparisons. This issue has implications for virtually all reported results and should be transparently addressed both in the abstract and in the main text.

The introduction effectively establishes the clinical relevance of pericarditis and highlights limitations of observational studies. However, the biological rationale linking metabolomics, and specifically bile acid metabolism, to pericarditis remains insufficiently developed. Because cholate ultimately emerges as the central mediator of the study, the reader would benefit from a more detailed discussion of bile acid signaling, immune regulation, and the gut-heart axis before entering the Methods section. At present, the cholate finding appears somewhat unexpected rather than hypothesis-driven. Additionally, the literature gap should be defined more precisely. Although the authors emphasize the need for causal inference, it would be valuable to explicitly state that no prior study has integrated immune-cell GWAS data, metabolomic GWAS resources, and mediation MR analysis in the setting of pericarditis.

The Methods section is generally comprehensive but would benefit from several additional analyses and clarifications. One concern involves the relatively small number of pericarditis cases available in the outcome GWAS, comprising only 1,795 affected individuals. While the total sample size is large because of the control population, the limited number of cases raises concerns regarding statistical power and the stability of effect estimates. A formal discussion of power considerations would strengthen confidence in the reported associations.

The instrumental variable selection strategy also warrants further justification. The use of a significance threshold of P < 1×10⁻⁵ for immune phenotypes and metabolites is less stringent than the conventional genome-wide significance threshold. While this approach is increasingly used in exploratory MR studies, it also increases the possibility of weak instruments and false-positive findings. The manuscript should clearly justify this choice and ideally provide sensitivity analyses using more conservative thresholds.

Another methodological limitation is the absence of colocalization analyses. Mendelian randomization alone cannot distinguish between a genuine causal effect and a scenario in which two traits are associated because of distinct variants in linkage disequilibrium. Given that the cholate-mediated pathway represents the principal mechanistic conclusion of the manuscript, performing colocalization analyses for this pathway would considerably increase confidence in the biological interpretation. Similarly, directionality analyses such as Steiger filtering should be considered to reinforce causal orientation.

The Results section presents a large number of statistically significant associations. However, the narrative could be improved by more clearly prioritizing the most robust findings. The identification of 29 immune phenotypes associated with pericarditis is potentially important, yet the sheer volume of results risks diluting the biological message of the study. Greater emphasis on the strongest associations, particularly those surviving multiple-testing correction, would enhance readability and impact.

The reverse MR section is interesting and adds depth to the analysis. Nonetheless, some conclusions appear overstated. The manuscript suggests that the absence of overlap between forward and reverse associations effectively rules out reverse causation. Although the findings certainly reduce concerns regarding reverse causality, MR cannot eliminate this possibility because several sources of bias may persist. More cautious wording would improve scientific rigor.

The mediation analysis is arguably the most novel component of the work but also the area most vulnerable to criticism. The rationale for selecting the CD25-IgD⁺CD38^bright^-cholate pathway from among several significant immune-metabolite associations should be clarified. Reviewers are likely to ask why this pathway was prioritized over alternative candidates identified in the screening analysis. Explicit objective criteria for pathway selection would therefore be valuable.

A further issue concerns the interpretation of effect directions. The reported beta coefficients indicate that increased CD25 expression on IgD⁺CD38^bright^ B cells is associated with increased pericarditis risk, while simultaneously reducing cholate levels, and cholate itself appears protective against pericarditis. However, portions of the manuscript subsequently describe this B-cell phenotype as protective. These statements appear internally inconsistent and should be carefully re-examined. Resolving this issue is essential because the central biological conclusion of the article depends upon the correct interpretation of effect directions.

The Discussion is generally thoughtful and integrates the findings with existing literature. Nevertheless, the language frequently implies greater certainty than warranted by the data. Expressions such as "proven causal driver," "first clear evidence," and "complete causal pathway" should be avoided. Mendelian randomization provides evidence that is consistent with causality but does not establish biological mechanisms in the same manner as experimental studies. Adopting more cautious language throughout the Discussion would improve credibility and align the manuscript with current standards in genetic epidemiology.

Finally, the limitations section is currently too brief relative to the complexity of the study. Beyond the acknowledged lack of experimental validation and the restriction to European ancestry populations, the authors should discuss the implications of multiple testing, the use of relaxed instrument-selection thresholds, the possibility of residual horizontal pleiotropy, the absence of colocalization analyses, and the relatively limited number of pericarditis cases in the outcome GWAS. A more comprehensive limitations section would demonstrate methodological awareness and likely pre-empt reviewer concerns.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Overall Assessment

This manuscript addresses a clinically relevant and potentially novel question by investigating whether circulating metabolites may mediate the relationship between immune-cell phenotypes and pericarditis. The integration of immune-cell, metabolomic, and pericarditis GWAS datasets is potentially valuable. However, several important issues concerning the statistical analysis, causal interpretation, and consistency of the proposed mediation pathway need to be addressed before the biological conclusions can be considered sufficiently supported.

In particular, the manuscript currently contains an inconsistency in the direction of the principal immune-cell–cholate–pericarditis pathway, while the statistical criteria used to identify the reported “causal immunophenotypes” do not appear to account adequately for multiple testing. Additional clarification of instrument strength, causal directionality, mediation methodology, and the biological interpretation of cholate is also required.

Major revision is therefore recommended.

Major Comments

1. Please reconcile the direction of the principal mediation pathway throughout the manuscript.

The direction of the reported CD25/IgD+ CD38br B-cell–cholate–pericarditis pathway appears inconsistent across different sections of the manuscript. Based on the reported estimates, the total effect of the immunophenotype on pericarditis is positive (β = 0.146; OR = 1.157). Furthermore, both the association between the immunophenotype and cholate (b1) and the association between cholate and pericarditis (b2) are negative. Thus, the proposed pathway appears to indicate that the immunophenotype increases pericarditis risk through a reduction in a protective metabolite.

However, the Abstract describes the pathway as reducing pericarditis risk by lowering cholate, which is inconsistent with the reported effect estimates. Similar inconsistencies appear elsewhere in the manuscript, including the statement that the phenotype maintains higher circulating cholate levels and the description of the pathway as protective.

The authors should carefully review the title, Abstract, Results, Discussion, and Figure 5 and ensure that the direction of the proposed pathway is presented consistently and correctly throughout.

2. The multiple-testing strategy for the initial screening requires substantial clarification and correction.

The analysis evaluates 731 immunophenotypes and approximately 1,400 metabolites using a nominal P < 0.05 threshold. Given the large number of comparisons, a substantial number of statistically significant associations would be expected by chance alone. Therefore, the identification of 29 immunophenotypes at this threshold does not, by itself, provide sufficient evidence to characterize these findings as “causal.”

A similar concern applies to the reverse-MR analysis and the reported absence of overlap between forward and reverse findings. The authors should clarify whether appropriate multiple-testing correction, such as false discovery rate or Bonferroni correction, was applied to these screening analyses. If not, the terminology describing the identified immunophenotypes as causal should be reconsidered, and the conclusions should be appropriately tempered.

3. The description of the P < 1 × 10⁻⁵ threshold as genome-wide significance should be corrected.

The manuscript refers to P < 1 × 10⁻⁵ as a genome-wide significance threshold. This terminology is inappropriate because the conventional genome-wide significance threshold is generally P < 5 × 10⁻⁸.

If the authors intentionally used P < 1 × 10⁻⁵ as a suggestive or instrument-selection threshold, this should be described accurately and justified. The manuscript should also acknowledge that use of a less stringent threshold may increase the likelihood of weak instruments and horizontal pleiotropy.

4. The strength and reliability of the reverse-MR analysis need to be demonstrated more clearly.

The pericarditis GWAS contains a relatively small number of cases compared with controls (1,795 cases versus 453,370 controls). This raises concerns regarding the availability and strength of genetic instruments for the reverse-MR analysis.

The authors should report the number of independent instruments retained for the reverse analysis, together with appropriate measures of instrument strength, including F-statistics where applicable. The interpretation of the reverse-MR findings as evidence of a “bidirectional architecture” should be moderated if the available instruments are few or weak.

In addition, the authors should provide further information regarding the definition of pericarditis in the underlying GWAS dataset, particularly because pericarditis may encompass heterogeneous etiologies with potentially different biological mechanisms.

5. The mechanistic interpretation of cholate should be more closely aligned with the available evidence.

The proposed biological mechanisms involving M2 macrophage polarization and TLR2 signaling appear to rely on evidence concerning tauroursodeoxycholic acid and ursodeoxycholic acid rather than cholate itself. These compounds should not be considered interchangeable without appropriate supporting evidence.

The authors should either provide evidence specifically supporting the proposed effects of cholate or clearly present these mechanisms as hypotheses extrapolated from related bile acids. The manuscript should also acknowledge potential sources of pleiotropy affecting circulating cholate, including hepatic, metabolic, dietary, and gut microbiome-related pathways. Where appropriate, potential pleiotropic effects of the genetic instruments should be evaluated.

6. Formal assessment of causal directionality should be incorporated.

Because the manuscript makes causal and mediation-related claims, an assessment of whether the genetic instruments preferentially explain variance in the exposure rather than the outcome would strengthen the analysis. The authors should consider applying Steiger filtering or an equivalent directionality test, particularly for the key immunophenotype-to-cholate and cholate-to-pericarditis relationships.

The results of this analysis should be reported and incorporated into the interpretation of the proposed causal pathway.

7. The methodology used to derive confidence intervals for the mediation estimates should be specified.

The indirect effect is based on the product of the two pathway coefficients (b1 × b2). The manuscript should explicitly describe the statistical method used to derive the corresponding 95% confidence intervals, including the confidence interval for the proportion mediated.

For example, if a delta-method or bootstrap approach was used, this should be stated clearly. Given the statistical properties of product-of-coefficients estimates, the choice of method can influence the resulting confidence interval and should therefore be transparent.

Minor Comments

1. Please clarify whether there was any participant overlap among the exposure, metabolite, and outcome GWAS datasets, as sample overlap can influence MR estimates.
2. The units used for the immunophenotype effects should be clarified. Because CD25 expression is measured using fluorescence intensity, reporting effects per standard deviation may provide greater interpretability than effects per raw unit.
3. Please clarify the unexpectedly low number of significant metabolites identified among approximately 1,400 metabolites using a nominal P < 0.05 threshold. This should be verified to ensure that all metabolites were included in the stated screening analysis and that the reported number is accurate.
4. The quoted statement in line 305 beginning with “Our MR findings give the first clear evidence...” appears stylistically inconsistent with the surrounding text. The quotation marks should be removed, and the wording should be moderated because MR analyses provide evidence consistent with causal relationships but do not independently establish biological mechanisms.
5. Since the analyses rely primarily on publicly available GWAS datasets, the Data Availability Statement should provide the relevant accession identifiers. Sharing the analysis code would also improve reproducibility.
6. Please verify the apparently duplicated CD33 immunophenotype entries in Table 1. If these represent distinct phenotypes, their definitions should be clearly differentiated; otherwise, the table should be corrected.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors addressed all concerns.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed the comments raised in the previous round. The instrument-selection strategy and the relaxed P-value thresholds are now clearly justified as exploratory, and the causal claims have been appropriately softened to reflect the hypothesis-generating nature of the analysis. The additional sensitivity analyses (MR-PRESSO, Steiger directionality, leave-one-out) and FDR correction adequately support the robustness of the key B cell–cholate–pericarditis pathway, and the limitations, small case number, single-ancestry data, absence of colocalisation and experimental validation, are candidly acknowledged.
I have no further concerns and recommend acceptance in its present form.