Review Reports

Comments 1: Lack of a clear central message.

It remains unclear whether the manuscript discusses: 1) Causal mechanisms of ASD, 2) Phenotypic modifiers, or 3) Biomarkers. This distinction must be explicitly clarified and consistently maintained throughout the manuscript.

Response 1: Thank you for this important and insightful comment. We agree that clarifying the central conceptual positioning of the manuscript is essential.

In response, we have substantially revised the manuscript to explicitly define and consistently maintain the scope of this review. Specifically, we now clearly position gut-derived metabolic imbalance-particularly short-chain fatty acids (SCFAs)-not as a primary causal mechanism or a diagnostic biomarker of ASD, but as a potential phenotypic modifier that may contribute to the observed clinical heterogeneity in a subset of individuals.

To address this point, we have made the following key revisions:

Introduction Section

- Clarification of the central framework: The manuscript now explicitly states that this review adopts a stratification-oriented and hypothesis-generating conceptual framework, rather than proposing a universal etiological model. We further clarify that the concept of “metabolic ASD” is not intended as a discrete clinical entity or biomarker-defined category, but as a conceptual framework referring to a putative subgroup characterized by context-dependent peripheral metabolic involvement.

- Removal of causal and deterministic interpretations throughout the manuscript: Terminology suggesting causality or definitive classification (e.g., “defined subtype,” “causal role”) has been systematically revised. These have been replaced with context-dependent and modifier-based expressions (e.g., “may contribute,” “phenotypic modifier,” “biologically plausible subgroups”) to ensure conceptual consistency.

- Explicit distinction from biomarkers and universal models: We have revised the manuscript to avoid framing SCFAs as diagnostic biomarkers. Instead, SCFAs are consistently described as context-dependent metabolic signals, and the overall framework is explicitly positioned as a hypothesis-generating, stratification-oriented approach rather than a biomarker-based classification.

Conclusion Section

- Reinforcement in the Conclusion: The revised Conclusion reiterates that gut-derived metabolic imbalance is framed as a phenotypic modifier contributing to ASD heterogeneity, rather than as a singular etiological determinant or diagnostic marker. We further emphasize that “metabolic ASD” represents a hypothesis-generating framework for stratified research, rather than a clinically established subtype.

Though these revisions, the manuscript now maintains a consistent and clearly defined conceptual position-focusing on modifier-based, context-dependent, and stratification-oriented interpretation of metabolic alternations in ASD-thereby addressing the reviewer’s concern regarding ambiguity in the central message.

We appreciate the reviewer’s comment, which has significantly improved the clarity and coherence of our manuscript.

Comments 2: “Metabolic ASD” is introduced, but it is not clearly explained.

It is unclear:

How would this subtype be identified?

What criteria would be used?

How does it differ from existing ways of testing done for ASD, and how can it be introduced in the protocol?

As a consequence, understanding how this concept could be applied in research or practice is challenging.

Response 2: We thank the reviewer for this important comment regarding the clarity and practical applicability of the “metabolic ASD” concept. We agree that additional clarification on how this framework may be interpreted and utilized in research contexts was necessary.

Importantly, our intention is not to define a clinically established subtype with fixed diagnostic criteria, but to propose a hypothesis-generating, stratification-oriented framework that can guide future research.

To address the reviewer’s concerns, we have revised the manuscript to clarify the operational implications of this framework as follows:

Introduction Section

- Conceptual identification approach: We now explicitly state that this framework does not rely on a single defining biomarker. Instead, it adopts a multi-dimensional approach integrating metabolic, immunological, and gastrointestinal features, which may collectively indicate context-dependent peripheral involvement.

Introduction Section & Section 3

- Criteria as integrative research dimensions: We clarify that no fixed diagnostic criteria currently define this subgroup. Rather, metabolic and microbial features are interpreted in combination, and we emphasize the need for stratified analytical approaches to capture biologically meaningful variability across individuals.

- Distinction from existing ASD evaluation frameworks: We emphasize that this framework does not replace established diagnostic approaches based on behavioral criteria. Instead, it provides an additional biological stratification perspective aimed at explaining heterogeneity rather than redefining diagnosis.

Conclusion Section

- Potential integration into research protocols: The revised manuscript now indicates that this framework may be incorporated into future studies through stratified cohort designs and multi-layer data integration, rather than as a standalone diagnostic classification.

These revisions clarify that “metabolic ASD” should be understood not as a fixed subtype, but as a conceptual and operational research framework that supports stratified, mechanism-driven investigation of ASD heterogeneity.

We appreciate the reviewer’s insightful comment, which has helped us strengthen the clarity and translational relevance of this concept.

Comments 3: Overly lengthy and unfocused

The manuscript is too lengthy and contains considerable repetition. Key messages are weakened by excessive mechanistic detail, which lowers clarity and readability. 

Response 3: We appreciate the reviewer’s thoughtful comment regarding the overall length, repetition, and clarity of the manuscript. In response, we have carefully revised the manuscript throughout to improve focus, reduce redundancy, and enhance readability, while preserving the scientific depth and integrity of the review.

Specifically, we implemented the following revisions:

- Reduction of repetition and improved conceptual clarity: Repetitive statements related to key themes (e.g., heterogeneity of ASD, context-dependent interpretation of SCFAs, and non-causal framing) have been consolidated to avoid redundancy and to ensure that the central message is presented more clearly and consistently.

- Streamlining of mechanistic descriptions: Sections containing excessive mechanistic detail have been carefully condensed. Detailed descriptions of molecular pathways and experimental findings have been simplified into concept-focused summaries, while retaining representative evidence and key references to preserve scientific rigor.

- Improved structural coherence and readability: Sentences and paragraphs have been revised for conciseness and logical flow. In particular, long or densely structured passages have been rephrased to improve readability and to better align with the integrative and conceptual nature of a review article.

- Balanced retention of quantitative and experimental evidence: While reducing verbosity, we ensured that essential quantitative findings and representative experimental evidence were retained, thereby maintaining the evidential strength of the manuscript.

Because these revisions were applied across multiple sections of the manuscript, specific line-by-line changes are not indicated here. However, the overall manuscript has been substantially refined to present a more concise, focused, and coherent narrative.

We believe that these revisions have significantly improved the clarity and readability of the manuscript while preserving its scientific contribution, and we sincerely appreciate the reviewer’s comment for helping us strengthen the overall quality of the work.

Comments 4: Insufficient differentiation between evidence types

Human, animal, and in vitro findings are often discussed collectively without a clear distinction. This can lead to overinterpretation of mechanistic findings. 

Response 4: We thank the reviewer for this important and insightful comment. We fully agree that insufficient differentiation between human, animal, and in vitro evidence may lead to overinterpretation of mechanistic findings.

In response, we have undertaken a comprehensive revision of the manuscript to clearly distinguish evidence types and to align the strength of interpretation with the level of evidence. Because this issue affects the overall narrative and multiple sections of the manuscript, the revisions were implemented systematically throughout the text rather than as isolated line-specific changes. Therefore, it is not feasible to indicate a limited number of specific lines where modifications were made.

The key revisions include:

- Explicit differentiation of evidence types: We have clarified whether findings are derived from human clinical studies, animal models, or in vitro systems throughout the manuscript. Where appropriate, we have explicitly indicated the study context (e.g., “clinical studies,” “in a mouse model,” “experimental systems,” “in vitro studies”) to improve transparency.

- Adjustment of claim strength: Mechanistic descriptions derived from preclinical studies have been revised to avoid causal or deterministic language. Expressions such as “demonstrated,” “induces,” or “leads to” have been replaced with more appropriate terms (e.g., “reported,” “suggests,” “may,” “is associated with”) to reflect the limitations of the evidence.

- Introduction of cross-level interpretation boundaries: Where findings from animal or in vitro studies are discussed in relation to human physiology or ASD, we have added explicit statements emphasizing the limitations of extrapolation. In particular, we have clarified that many mechanistic insights are derived from experimental systems and may not directly translate to human neurodevelopmental contexts.

- Reframing of mechanistic interpretation: Throughout the manuscript, we have revised the narrative to position gut-derived metabolites, including SCFAs and propionate, as context-dependent modulators rather than direct causal determinants. This framing ensures that mechanistic findings are interpreted within an appropriate biological and methodological context.

- Strengthening of limitation-focused sections: We have expanded and refined Sections 5.4 and 6.2 to explicitly address the limitations of preclinical evidence and to define appropriate methodological requirements for future research. These sections reinforce the distinction between exploratory mechanistic insights and clinically validated evidence.

Collectively, these revisions ensure a clearer separation of evidence levels and substantially reduce the risk of overinterpretation. We believe that the revised manuscript now more accurately reflects the current state of the evidence and provides a balanced and methodologically sound interpretation of the literature.

Comments 5: The developmental window is not clearly defined.

When discussing critical windows, it remains unclear whether the authors refer to the prenatal or postnatal period. The manuscript mentions various stages of development but does not explicitly clarify which periods are most significant, how they differ, or how they are specifically linked to ASD. Consequently, the importance of timing in the proposed mechanisms remains ambiguous.

Response 5: We thank the reviewer for this important comment. We agree that the initial version of the manuscript did not clearly define developmental windows or sufficiently distinguish between prenatal and postnatal periods, which may have led to ambiguity regarding the timing and relevance of the proposed mechanisms.  

In response, we have revised the manuscript to explicitly define and structurally organize developmental windows, as well as clarify their differential relevance to ASD.

Specifically, the following revisions were made:

- Explicit definition of developmental windows: In Section 2.4, we have introduced a clear conceptual framework distinguishing developmental stages into prenatal (maternal–fetal), early postnatal (infancy), and later developmental periods. This addition provides a structured temporal framework for interpreting SCFA-related effects.

- Clarification of stage-specific biological characteristics: We have expanded the description of each developmental period to highlight differences in neurodevelopmental processes, microbiota establishment, and immune maturation. In particular, we emphasize that the prenatal and early postnatal periods represent phases of heightened sensitivity to metabolic and immune signals.

- Strengthened linkage to ASD vulnerability: We have clarified how these developmental windows relate to ASD by explicitly stating that early developmental stages coincide with critical processes such as synaptogenesis, microglial maturation, and neural circuit formation, which may confer increased susceptibility to metabolic perturbations.

- Reinforcement in the Conclusion section: To ensure conceptual continuity, we have incorporated a concise statement in the Conclusion highlighting that metabolic influences may be particularly relevant during prenatal and early postnatal developmental windows.

Collectively, these revisions clarify the temporal dimension of the proposed framework and reduce ambiguity regarding the role of timing in gut–brain axis–related mechanisms. We believe that the revised manuscript now provides a clearer and more structured interpretation of developmental sensitivity in ASD.

Comments 6: Lack of integration with current ASD models

The manuscript does not adequately engage with polygenic risk score approaches and gene–environment interaction models. Existing ASD models that incorporate both genetic risk (e.g., polygenic risk scores) and environmental influences, along with their interactions, are not adequately discussed. This limits its relevance within the current understanding of ASD aetiology. 

Response 6: We thank reviewer for this important and constructive comment. We agree that clearer integration with contemporary ASD models, particularly those incorporating polygenic risk scores (PRS) and gene-environment interactions, is essential for situating our framework within the current understanding of ASD aetiology.

In response, we have revised the Introduction (section 1.1) to explicitly incorporate recent advances in PRS-based approaches and gene-environment interaction frameworks. Specifically, we have added statements emphasizing that ASD arises from the cumulative effects of multiple genetic variants and that genetic susceptibility operates in interaction with environmental and biological factors rather than in isolation.

In addition, we have clarified how the proposed “metabolic ASD” framework aligns with and extends these models. We now explicitly position peripheral metabolic and immune environments as a potential modulatory layer through which environmental influences may interact with underlying genetic vulnerability. This integration highlights that our framework is not intended to replace existing genetic or gene-environment models, but rather to complement them by providing a biologically plausible intermediate layer linking environmental exposures to neurodevelopmental outcomes.

Collectively, these revisions strengthen the conceptual linkage between gut-derived metabolic mechanisms and established ASD aetiological models. We believe that the revised manuscript now more clearly reflects the current multi-layered understanding of ASD, in which genetic risk, environmental influences, and their interactions jointly contribute to phenotypic heterogeneity.

Comments 7: The brain-centric model is not sufficiently clarified

The manuscript criticises a “brain-centric” view without adequately defining or critically examining it. A clearer explanation of this model is necessary to support the proposed alternative. 

Response 7: We thank the reviewer for this insightful comment. We agree that the initial version of the manuscript did not clearly define what is meant by a “brain-centered” model, which may have limited the clarity of our conceptual argument.

In response, we have revised the Introduction (Section 1.1) to explicitly define and contextualize brain-centered models. Specifically, we now describe these models as frameworks that primarily attribute ASD pathophysiology to intrinsic abnormalities within the central nervous system, including synaptic dysfunction, altered neural connectivity, and disrupted neurodevelopmental signaling pathways.

In addition, we have clarified the scope and limitations of these models. While acknowledging that brain-centered approaches have provided critical insights into ASD, we now explicitly state that they may not fully account for the observed clinical heterogeneity or the contribution of peripheral biological systems, such as metabolic, immune, and gastrointestinal processes.

These revisions allow us to more clearly position our proposed “metabolic ASD” framework not as a replacement for brain-centered models, but as a complementary extension that incorporates interactions between central and peripheral systems. By doing so, we aim to provide a more integrative conceptual framework that better reflects the multi-layered nature of ASD pathophysiology.

We believe that these changes improve the clarity and conceptual coherence of the manuscript and strengthen the rationale for the proposed framework.

Comments 8: Excessive use of conceptual terminology

The manuscript frequently depends on abstract terms (e.g., “conceptual framework”) without offering clear, testable definitions or practical examples. 

Response 8: We thank the reviewer for this constructive comment. We agree that the initial version of the manuscript relied heavily on conceptual terminology without sufficiently clarifying how the proposed framework can be operationalized or tested.

In response, we have revised the manuscript to strengthen the linkage between the conceptual framework and its practical and testable applications.

First, in the Introduction (Section 1.1), we have clarified the operational meaning of the proposed “metabolic ASD” framework by explicitly stating that it can be translated into testable research strategies through the integration of metabolic, immune, and gastrointestinal biomarkers with clinical phenotyping. This addition provides a concrete basis for how the framework can be applied in empirical research.

Second, in Section 6.2, we have incorporated a specific example of how the framework may be operationalized in practice. In particular, we now describe longitudinal cohort designs that stratify individuals based on metabolic and gastrointestinal profiles and examine their association with neurodevelopmental trajectories. We further emphasize that such approaches enable the formulation of testable hypotheses regarding whether specific metabolic profiles precede or modify ASD-related phenotypes.

These revisions reduce reliance on abstract terminology and provide clearer guidance on how the proposed framework can be translated into measurable, hypothesis-driven research. We believe that the manuscript now more effectively balances conceptual integration with practical applicability.

Comments 9: Need for schematic figures

Given the complexity of the described mechanisms, at least one comprehensive schematic diagram is essential to enhance readability and conceptual clarity.

Response 9: We thank the reviewer for this helpful suggestion. We agree that a comprehensive schematic representation is important to enhance readability and conceptual clarity given the complexity of the mechanisms discussed.

In response, we have included a new schematic diagram (Figure 2) that summarizes the overall conceptual framework of the manuscript. This figure integrates key elements of the proposed model, including gut microbiota–derived metabolite imbalance, peripheral metabolic and immune processes, gut–brain axis signaling pathways, and their potential influence on neurodevelopmental outcomes in ASD.

To further improve conceptual integration, we have revised the figure to explicitly incorporate genetic susceptibility and environmental inputs (e.g., polygenic risk and environmental factors), thereby aligning the schematic with current gene–environment interaction models of ASD.

In addition, the figure legend has been revised to reduce reliance on abstract terminology and to emphasize the testable nature of the proposed framework, including how the model may generate hypothesis-driven research.

Finally, we have strengthened the linkage between the schematic and the main text by explicitly referencing Figure 2 in the manuscript (Section 6.3), where it is used to illustrate how genetic, environmental, and metabolic factors are integrated into a unified and testable framework of ASD heterogeneity.

We believe that these revisions substantially improve the clarity, accessibility, and integrative presentation of the manuscript.

Comments 10: Repetition and redundancy

Several concepts (e.g., SCFA heterogeneity, the role of propionate) are repeated across sections and should be streamlined. 

Response 10: We thank the reviewer for this valuable comment. We agree that certain concepts, including SCFA heterogeneity and the role of propionate, were reiterated across multiple sections in the original version of the manuscript, which could reduce clarity and conciseness.

In response, we have revised the manuscript to streamline the presentation and reduce redundancy, particularly in Section 6. Rather than repeating detailed descriptions, we have clarified the distinct role of each section by confining (i) empirical observations of SCFA heterogeneity primarily to Section 3 and (ii) mechanistic discussions of propionate to Section 4. In Section 6, these concepts are now referenced concisely using cross-references (e.g., “as summarized in Section 3” and “as discussed in Section 4”) and are interpreted at a higher level, focusing on their implications for clinical heterogeneity, study design, and therapeutic strategy.

In addition, several sentences in Section 6 have been revised to improve conciseness and eliminate redundant phrasing, thereby enhancing the overall logical flow and readability of the manuscript.

We believe that these revisions have substantially reduced repetition while preserving the necessary conceptual continuity across sections.

Comments x: Minor Issues

·         Acronyms should be defined once and used consistently thereafter.

·         Some references should be more carefully selected and better contextualised, particularly when supporting broad mechanistic claims (e.g. ref 1) 

Response x: We thank the reviewer for these helpful comments.

First, all acronyms have been carefully reviewed, defined at first mention, and used consistently throughout the manuscript.

Second, while the core reference set has been retained, we have carefully revised how references are presented and interpreted throughout the manuscript. In particular, for statements involving broad mechanistic implications, we have refined the wording to better reflect the level of evidence and added appropriate contextualization (e.g., distinguishing experimental, animal, and clinical findings). These revisions ensure that the cited references more accurately support the claims made without overstating their implications.

We believe that these changes improve the clarity, consistency, and scientific rigor of the manuscript.

4. Response to Comments on the Quality of English Language

Response 1: We thank the reviewer for the positive assessment of the language quality. No major revisions were required; however, minor edits have been made to further improve clarity and consistency. 

5. Additional clarifications

We would like to emphasize that the revised manuscript has been carefully updated to improve overall clarity, conceptual coherence, and consistency across sections. In particular, we have strengthened the integration between conceptual frameworks and testable research directions, clarified distinctions between different levels of evidence, and enhanced the linkage between the main text and the schematic illustration (Figure 2). We believe that these revisions have substantially improved the overall quality and readability of the manuscript.

Comments 1:

The manuscript proposes a conceptual framework termed ‘metabolic ASD’, identifying a subgroup of individuals with Autism Spectrum Disorder (ASD) in whom peripheral metabolic imbalances, mediated by gut microbiota dysbiosis, alterations in short-chain fatty acids (SCFAs) and mitochondrial dysfunction, may contribute to neurodevelopmental vulnerability. The argumentative structure is coherent and the topic is of current scientific relevance. The authors demonstrate a broad knowledge of the literature and correctly adopt an epistemically cautious tone, avoiding premature causal claims. However, the manuscript presents some structural, methodological and presentation issues which, in the reviewer’s opinion, must be addressed before the work can be considered for publication.

Response 1: We sincerely thank the reviewer for the constructive and encouraging evaluation of our manuscript, and for recognizing the conceptual relevance, overall coherence, and appropriately cautious interpretation of the proposed framework.

We fully agree with the reviewer’s assessment that further improvements in structure, methodological clarity, and overall presentation are necessary to strengthen the manuscript. Accordingly, we have carefully revised the manuscript throughout to address these points.

Specifically, we have (1) refined the overall structure to improve the logical flow between sections, (2) clarified the conceptual and methodological positioning of the “metabolic ASD” framework to reduce potential ambiguity, and (3) revised the language and presentation to enhance clarity, precision, and readability.

As these revisions were applied across multiple sections, the changes have been implemented consistently throughout the manuscript rather than being limited to a single location.

Comments 2:

The ‘Conclusions’ section is numbered ‘5’ in the main text, but the preceding sections go up to 6 (Section 6.3). The authors should review and standardise all section numbering.

Response 2: We thank the reviewer for pointing out this inconsistency. The section numbering has been corrected, and the “Conclusions” section has been appropriately renumbered to ensure consistency with the preceding sections. In addition, we have carefully reviewed the entire manuscript to ensure that all section numbering is accurate and consistent throughout.

Comments 3:

Table 1, ‘Reported alterations in SCFA levels in ASD’, is incomplete: it lists only two studies (Wang et al., 2019; Liu et al., 2019), yet the text cites a broader body of literature. This is the key table supporting the argument of clinical heterogeneity. I recommend that the authors expand the table to include all available quantitative studies with data on acetate, propionate and butyrate, specifying for each: sample type, age, number of patients, number of controls, analytical method and units of measurement.

Response 3: We thank the reviewer for this important and constructive comment. We agree that Table 1 plays a key role in supporting the interpretation of clinical heterogeneity in SCFAs alterations in ASD.

In response, we have substantially expanded Table 1 to include additional human case–control studies reporting quantitative data on acetate, propionate, and butyrate. For each study, we have included detailed information on sample type, age group, number of participants (ASD and control), and analytical methods, in line with the reviewer’s recommendation.

To maintain clarity and readability, Table 1 presents representative studies that illustrate the heterogeneity of reported SCFAs alterations across cohorts. In addition, we have revised the main text (Section 3.1 and Section 3.2) to explicitly reference Table 1 and to clarify that these findings reflect substantial inter-study variability, while acknowledging that additional studies further support this heterogeneity but differ in design and reporting formats, limiting direct comparability.

We believe that these revisions have significantly strengthened the presentation and interpretation of SCFAs-related clinical variability in ASD.

Comments 4:

I would advise the authors to expand on the scientific background in the introduction, perhaps also describing natural substances capable of improving the gut microbiota and, consequently, the autism spectrum. I recommend consulting: DOI: 10.3390/nu17060948. 

Response 4: We thank the reviewer for this helpful suggestion and for pointing out the relevance of microbiota-modulating approaches, including dietary factors.

In response, we have expanded the Introduction (Section 1.2) to briefly acknowledge that dietary patterns, including microbiota-modulating diets such as the Mediterranean diet, may influence gut microbial composition and metabolic outputs. To maintain the focus and scope of the present review, this addition was intentionally kept concise and framed within the context of gut microbiota–derived metabolic modulation rather than as a detailed discussion of therapeutic interventions.

In addition, we have incorporated the reviewer-recommended reference and further addressed this point in Section 6.3, where we discuss dietary and microbiota-targeted strategies within the broader framework of context-dependent and subgroup-specific interventions. This allows us to acknowledge the potential relevance of such approaches while preserving the primary objective of the manuscript, which is to provide a mechanistic and conceptual framework for understanding metabolic heterogeneity in ASD.

We believe that these revisions appropriately address the reviewer’s suggestion while maintaining the conceptual focus and coherence of the manuscript.

Comments 5:

The concept of ‘metabolic ASD’ is never rigorously operationalised. The authors propose a list of candidate biomarkers (Table 5) but do not define thresholds or inclusion/exclusion criteria, nor do they provide an estimated prevalence of this subtype within the general ASD population. How does one clinically distinguish a patient with ‘metabolic ASD’ from one without? Without a minimum operational definition, the concept remains purely speculative. It is recommended that a subsection be added dedicated to the proposed operational definition, even at a hypothetical level, or that the proposal be explicitly redefined as purely conceptual. 

Response 5: We thank the reviewer for this important and insightful comment regarding the operationalization of the proposed “metabolic ASD” framework. We agree that the concept, as initially presented, may appear insufficiently defined for clinical application. We would like to clarify that “metabolic ASD” is not intended as a formally defined clinical subtype, but rather as a hypothesis-generating conceptual framework aimed at guiding the identification of biologically meaningful subgroups within the heterogeneous ASD population.

In response to this comment, we have added a new subsection (Section 6.4, “Toward an Operational Framework for ‘Metabolic ASD’”) to explicitly address this issue. In this section, we clarify the current limitations of the concept, including the absence of validated thresholds, inclusion/exclusion criteria, and prevalence estimates. At the same time, we outline potential domains that may contribute to future operationalization, such as gut-derived metabolic profiles, gastrointestinal symptoms, markers of barrier integrity and inflammation, and broader metabolic context.

Importantly, we emphasize that any future distinction between individuals with and without “metabolic ASD” is unlikely to rely on single biomarkers or fixed cut-offs, but rather on the convergence of multiple biological signals within a stratified framework. Accordingly, the concept is explicitly positioned as a research-oriented framework, not as a diagnostic classification.

We believe that this addition clarifies the scope, limitations, and intended application of the “metabolic ASD” concept, and appropriately addresses the reviewer’s concern.

Comments 6:

The cited data on mitochondrial dysfunction (e.g. reduced ETC complexes, increased ROS) derive almost exclusively from animal models involving intracerebroventricular administration of propionate, from cell lines, or from patients with propionic aciduria — a metabolic condition that is genetically distinct from ASD. The inferential leap between these models and the ASD population is not sufficiently justified. The authors should either (i) add an explicit paragraph on the limitations of this extrapolation, or (ii) include direct human data on mitochondrial dysfunction in ASD not related to propionic aciduria (e.g. Giulivi et al. 2010, cited above), integrating them more systematically into the reasoning. 

Response 6: We thank the reviewer for this important comment regarding the interpretation of mitochondrial dysfunction and the potential overgeneralization from experimental models to the ASD population.

We agree that much of the evidence linking propionate exposure to mitochondrial dysfunction is derived from experimental systems, including animal models, in vitro studies, and investigations of propionic aciduria, which is a genetically distinct metabolic disorder. To address this concern, we have revised Section 4.3 to explicitly clarify the limitations of extrapolating these findings to ASD. Specifically, we now emphasize that such evidence should be interpreted as illustrative mechanistic models and does not establish a direct causal relationship between propionate and mitochondrial dysfunction in individuals with ASD.

In addition, we have incorporated human-based evidence of mitochondrial alterations in ASD independent of propionic aciduria. This includes studies such as Giulivi et al. (2010) and related reports demonstrating mitochondrial dysfunction, altered respiratory chain activity, and redox imbalance in subsets of individuals with ASD. Importantly, we have explicitly highlighted the heterogeneity of these findings, noting that mitochondrial dysfunction is not uniformly observed across all ASD cohorts.

Finally, we have revised the overall tone of the section to avoid causal overinterpretation and to frame mitochondrial dysfunction as a context-dependent and subgroup-specific feature rather than a universal or etiological mechanism.

We believe that these revisions address the reviewer’s concern by providing a more balanced and evidence-based interpretation of the available data.

Comments 7:

The section on the limitations of preclinical evidence is included (5.4) but is relatively brief and unsystematic. There is no discussion of: (i) the issue of dose-relevance (the concentrations of propionate used in animal models often exceed human physiological levels by orders of magnitude), (ii) publication bias in studies on the microbiota and ASD, (iii) the issue of reverse causality (changes in the microbiota may be a consequence of the selective eating behaviours typical of ASD, rather than a cause). It is recommended that this section be expanded or that these limitations be incorporated into the relevant sections. 

Response 7: Thank you for this important and constructive comment. We agree that the original limitation section (Section 5.4) was relatively brief and did not sufficiently address several key methodological and interpretative issues.

In response, we have revised and expanded Section 5.4 to provide a more structured and systematic discussion of current limitations. Specifically, we have incorporated three key aspects highlighted by the reviewer:

(i) Dose relevance: We now explicitly note that many preclinical studies employ propionate concentrations that exceed physiologically observed levels in humans, which limits the direct translational relevance of these findings.

(ii) Publication bias: We have added a discussion addressing the potential overrepresentation of positive findings in microbiota–ASD research, which may lead to an overestimation of the consistency and strength of reported associations.

(iii) Reverse causality: We have included a discussion emphasizing that alterations in gut microbiota may, at least in part, reflect behavioral characteristics of ASD (e.g., selective eating patterns), rather than serving as primary causal drivers.

These points have been incorporated into a revised and more clearly structured Section 5.4 with each limitation presented as a distinct conceptual component. In addition, these considerations have been conceptually aligned with the interpretation of evidence throughout the manuscript to ensure a more cautious and balanced framing.

We believe that these revisions strengthen the critical appraisal of the current literature and improve the overall interpretative rigor of the manuscript.

Comments 8:

Figure 1 presents a visually effective conceptual model, but the description in the caption uses language that is at times more assertive than the main text (e.g. ‘SCFAs... act as a context-dependent pathophysiological contributor’). Furthermore, the arrow from panel B (Metabolite & Barrier Dysregulation) to panel C (Gut-Brain Axis Pathways) suggests a linear causal sequence that the text itself contradicts. The authors should revise the figure to make it consistent with the epistemic tone of the manuscript, adding notes that make explicit the hypothetical nature of the proposed model. 

Response 8: We thank the reviewer for this insightful and constructive comment. We agree that the original version of Figure 2 (Figure 1 → Figure 2) and its caption could be interpreted as more assertive than the main text and that certain visual elements might suggest a linear causal sequence.

In response, we have revised both the figure and its legend to ensure consistency with the epistemic tone of the manuscript and to more clearly reflect the conceptual and hypothesis-generating nature of the proposed framework.

Specifically, we have made the following changes:

- Caption tone adjustment: The figure legend has been revised to reduce assertive language and to align with the cautious interpretative framework used throughout the manuscript. Terms implying direct causality have been replaced with expressions such as “may,” “associated with,” and “proposed,” to reflect the current level of evidence.

- Removal of implied linear causality: Visual elements that could be interpreted as indicating a unidirectional causal sequence have been modified. In particular, the connection between panels B and C has been changed to a non-directional dashed line to indicate a proposed association rather than a direct causal pathway. Additional directional cues that could imply linear progression have also been minimized.

- Explicit clarification of the model as conceptual and hypothetical: The revised legend now explicitly states that the figure represents a conceptual and hypothesis-generating model. It also clarifies that the connections shown represent context-dependent associations rather than definitive causal relationships.

These revisions were made to ensure that the figure does not overstate mechanistic conclusions and remains fully consistent with the cautious and integrative interpretation presented in the main text.

We believe that the revised figure now more accurately conveys the intended conceptual framework and addresses the reviewer’s concerns.

Comments 9:

I would advise the authors to include a few more images. Visual elements are essential for such a complex and structured review.

Response 9: We thank the reviewer for this helpful suggestion. We agree that visual elements can significantly enhance the clarity and accessibility of complex and structured content.

In response, we have added an additional schematic figure (Figure 1) to provide a concise summary of the proposed mechanisms discussed in Section 4, with a particular focus on propionate. This figure is intended to complement Figure 2, which presents the overall conceptual framework, by offering a more detailed and focused visualization of the context-dependent biological processes described in the text.

We believe that the inclusion of this additional figure improves the overall readability of the manuscript and facilitates a clearer understanding of the mechanistic relationships discussed.

4. Response to Comments on the Quality of English Language

Response 1: We thank the reviewer for this helpful comment. We have carefully revised the manuscript to improve clarity, readability, and overall language quality. In particular, we have refined sentence structure, improved terminology consistency, and adjusted phrasing to better reflect the intended scientific meaning. We believe that these revisions have enhanced the clarity and precision of the manuscript. 

5. Additional clarifications

We would like to take this opportunity to clarify the overall interpretative framework of the present review. Throughout the revision process, we have carefully refined the manuscript to ensure that all proposed mechanisms, particularly those related to gut-derived metabolites such as propionate, are presented within a cautious and context-dependent framework.

In particular, we emphasize that the concept of a “metabolic subtype” of ASD is intended as a hypothesis-generating perspective rather than a formally established biological classification. Similarly, the mechanistic pathways discussed are not presented as direct causal relationships, but rather as potential and context-dependent interactions supported primarily by preclinical and emerging clinical evidence.

We have also strengthened the discussion of methodological and interpretative limitations, including issues related to dose relevance, publication bias, and reverse causality, to provide a more balanced and critical appraisal of the current literature.

We hope that these revisions clarify the scope and intent of our work and contribute to a more nuanced understanding of the complex and heterogeneous nature of ASD.