JCI Response to Reviewers: A Revision Guide for Translational Evidence

Treat each theme as one link in the translational evidence chain. A local reply is not complete until the corresponding mechanism, cohort, and disease-facing statements agree across the response and manuscript.

Causal mechanism

When the reviewer asks whether a pathway is causal, more expression data rarely resolves the concern. Determine which link is missing: necessity, sufficiency, ordering, cell specificity, rescue, or independence from an alternative pathway.

If the revised experiment closes only one link, say so. Replace broad verbs such as "drives" or "mediates" when the evidence supports association, contribution, or requirement under a defined condition. Update the abstract and model figure, not only the detailed Results text.

Human or disease-facing validation

A human cohort can support relevance without proving the same causal mechanism established in a model system. Keep those evidentiary roles separate. State whether the cohort replicates direction, associates with disease severity, predicts an outcome, or shows the pathway in relevant tissue.

Report inclusion criteria, treatment context, disease stage, missing data, and major confounders. If the available cohort cannot answer the request, limit the inference by population and explain what future validation remains necessary.

Statistics and cohort structure

For observational or clinical data, a robust response identifies the model, covariates, endpoint definition, missing-data handling, multiplicity approach, and sensitivity analyses. Report effect estimates and confidence intervals where appropriate, not only threshold language.

When a result weakens after the correct analysis, disclose that directly and revise the conclusion. A smaller, defensible translational claim is more credible than preserving the original wording around a changed estimate.

Clinical relevance

Avoid turning biological relevance into an unsupported treatment recommendation. Explain whether the revised evidence identifies a target, stratification signal, disease mechanism, or testable clinical hypothesis. If no prospective clinical utility was evaluated, say so.

The JCI response should help the editor see why the work matters to human disease while preserving the line between mechanistic insight and demonstrated clinical action.

Disagreement works when the response improves the paper even if the exact request is declined. Clarify the design, add the missing analysis, or narrow the claim.

Use a decision sequence rather than a reflexive "beyond scope" answer:

1. Name the scientific uncertainty behind the request.
2. State whether the paper's current claim depends on resolving it.
3. Explain why the proposed study is unavailable or not the best test.
4. Provide an alternative analysis or evidence when valid.
5. Narrow the claim and add the unresolved issue as a limitation.

For example, if an independent prospective cohort cannot be assembled during revision, do not imply that a retrospective sensitivity analysis is equivalent. Explain what the sensitivity analysis does establish, limit the population claim, and avoid clinical-prediction language.

A response to a major request should state design, result, interpretation, and manuscript consequence.

Reviewer 2, Comment 4: The proposed disease pathway is not supported in human tissue and may be specific to the experimental model.

Response: We agree that the original manuscript did not establish the pathway's presence in human disease. We added immunostaining and transcript analysis in an independent tissue cohort, using prespecified inclusion criteria and blinded quantification. Both measures showed the predicted direction, with effect estimates and confidence intervals reported in new Figure 6 and Table S4. The cohort supports disease relevance but does not prove the causal sequence established in the model. We revised the abstract on page 2, lines 36-40, the Results on page 12, lines 338-372, and the limitation on page 16, lines 468-482.

The boundary sentence prevents the new cohort from being used to claim more than it tested.

JCI revisions can touch human-subject reporting, experimental methods, statistics, figures, and data availability. Reconcile these surfaces before upload:

• sample sizes, exclusions, and cohort labels across the flow diagram, tables, Methods, and supplement
• primary and secondary endpoints across the response and manuscript
• treatment exposure and covariates across models
• effect sizes, intervals, and p-values across text and figures
• ethics approval and consent language against the actual study design
• data and code availability against repository access
• disease and clinical claims against the limitations conceded in the response

Generate page and line references only after the clean manuscript is stable.

Across our JCI pre-submission and revision reviews, the strongest responses treat translation as an evidence chain rather than a rhetorical label. Three patterns repeatedly decide whether the revised paper feels coherent. These are Manusights editorial observations, not official JCI acceptance criteria.

The JCI mechanism and patient result answer different questions

The model-system experiment may establish necessity while the patient cohort establishes association. A weak response merges the two and claims that the human data validate the mechanism. We test each JCI figure and paragraph for its exact evidentiary role. The response should tell the editor whether the new experiment changes causal confidence, disease relevance, population generality, or clinical interpretation. When those roles are explicit, the abstract and Discussion become more precise and the reviewer can evaluate each link on its own terms.

The JCI cohort correction is hidden in the supplement

Reviewers often identify confounding, treatment imbalance, missingness, or selection criteria that materially change the result. Authors may run the correct sensitivity analysis but bury it in a supplement and preserve the original headline. We compare the revised cohort table, Methods, main result, abstract, and limitation. If the adjusted estimate is weaker, the main text must show it. A JCI response earns trust by explaining how the conclusion changed, not by using supplementary placement to minimize the change.

The JCI revision keeps a clinical verb the study did not test

Words such as "predicts," "improves," "guides," or "supports treatment" carry design requirements. In our review work, these verbs often survive even after authors concede that the study is retrospective, exploratory, or not externally validated. We audit the title, abstract, figure captions, and final Discussion paragraph after every major concession. Replacing a clinical-action claim with a disease-mechanism or hypothesis-generating claim does not weaken a good paper; it aligns the statement with the evidence the JCI reviewer can actually verify.

These failures usually appear when authors optimize each reply separately instead of checking whether the complete revised paper now supports one bounded translational conclusion.

The response confuses added volume with added evidence

More samples, stains, or correlations do not automatically resolve a causal concern. Explain how the new design distinguishes alternatives.

A disagreement never changes the manuscript

Even when the reviewer is not asking for the right experiment, the comment often reveals unclear wording or a missing limitation. Make that improvement visible.

Human data become a prestige signal

Including patient material is not enough. Define what the cohort contributes and avoid using human evidence to imply clinical utility that was not evaluated.

The response omits negative or changed results

Report when a requested analysis does not support the original expectation. Revise the claim and explain what remains informative.

Most major revisions are still conditional, and rejection remains possible when the causal mechanism is not closed, the human evidence is overinterpreted, or a corrected analysis materially weakens the main claim without a corresponding rewrite. A long response letter does not compensate for a broken evidence chain.

Think twice before resubmitting if the decision requires human validation that is central to the translational claim and the revision has no valid human evidence or defensible claim boundary. The better choice may be more work or a venue whose scope matches the evidence already available.

• The editor's controlling issues are summarized before the reviewer-by-reviewer replies.
• Every comment is reproduced and answered.
• Every change has a verified page, line, figure, table, or Methods reference.
• Mechanistic and human evidence are described according to their actual roles.
• Cohort definitions, exclusions, endpoints, and models match across all files.
• Clinical language does not exceed the design.
• Reviewer comments and author replies are visually distinct.
• The clean manuscript, marked copy, response, figures, tables, and supplement are final and mutually consistent.

Use the free revision readiness scan to test whether the revised evidence, not only the prose, answers the decision letter.

We reviewed current JCI author and publication-ethics materials, established response-letter guidance, and recurring translational-evidence problems from Manusights revision work. Use this page when a JCI revision must connect model-system mechanism, human relevance, cohort analysis, and the final disease-facing claim.

Reviewed July 12, 2026.

• JCI publication ethics and peer-review process
• JCI information for authors
• PLOS Computational Biology: Ten Simple Rules for Writing a Response to Reviewers
• Nature Computational Science guidance on responding to peer review

Official JCI sources support the journal's peer-review and author requirements. The response matrices and failure patterns are Manusights editorial guidance and should not be read as a guarantee of a JCI decision.