Longitudinal Fecal Short-Chain Fatty Acid Trajectories in Preterm Infants with Early-Onset Neonatal Sepsis: A Pilot Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Abstract, line 19: The phrase " a broad redistribution of the fecal SCFA pool" is not clear and should be omitted and the following sentence should follow after the word "exhibited". The result section needs to report the number of infants included and the average time point of onset of sepsis.

Introduction: There is no explanation in the introduction why an entire month of fecal SCFA analysis was chosen to relate to early onset sepsis.

Methods: It is not justifiable to exclude the NEC-risk subgroup from the between-group comparisons. Table 1 is a result and should be in the result section. Table 1 need to include type antibiotic use and timing of first antibiotic use in both groups as antibiotics are often given in preterm babies at birth and may influence the gut microbiome. The timing of the first stool sample on day 3 will have been after onset of early onset sepsis and hence first antibiotic administration in the majority of patients. Stool profiles therefore reflect result of sepsis and result of treatment intervention and not predisposition. The authors need to carefully and precisely report and analyse exact timing of first antibiotic administration and first stool sampling as well as all data on inflammatory markers and their timing after birth.

There is no sample size calculation presented. The study was underpowered to investigate NEC.

Author Response

Dear Reviewer,

We greatly appreciate your careful reading of our manuscript and your thoughtful suggestions. Your comments have helped us improve the clarity and quality of the work. Below we provide a point-by-point response, and all revisions have been highlighted in the manuscript.

 

Comment 1. The phrase ‘a broad redistribution of the fecal SCFA pool’ is not clear and should be omitted and the following sentence should follow after the word ‘exhibited’.

Response 1. Thank you for the comment. We have removed the unclear phrase and revised the sentence accordingly. The description now directly follows the word “exhibited,” as suggested (highlighted in the manuscript):

“At the earliest sampling window (TP1), infants with EONS exhibited a significantly lower median relative fraction of acetic acid (86.6% vs. 94.5% in non-sepsis), while several non-acetate components were relatively enriched.”

 

Comment 2. The result section needs to report the number of infants included and the average time point of onset of sepsis.

Response 2. Thank you for the comment. We have added the number of infants included in the study (49 in total; 18 with EONS and 31 without sepsis) to the beginning of the Results section. We have also added information on the timing of onset of sepsis. As described in the revised manuscript, all neonates with EONS demonstrated clinical suspicion within the first 6 hours of life, and the diagnosis was confirmed by laboratory evaluation within the first 72 hours. These additions are now clearly reported and highlighted in the Results section.

 

Comment 3. Introduction: There is no explanation in the introduction why an entire month of fecal SCFA analysis was chosen to relate to early onset sepsis.

Response 3. Thank you for this valuable comment. We have now added a clear justification in the “Introduction” section explaining why SCFA profiling was performed across the entire first postnatal month. As outlined in the revised manuscript, this period represents the most dynamic phase of intestinal and microbial maturation in preterm infants, during which early systemic events such as sepsis and antibiotic exposure can have prolonged metabolic consequences. We also incorporated supporting literature demonstrating week-to-week shifts in SCFA production and in other fecal biomarkers, which further validates the longitudinal design. The new text has been added to the end of the Introduction and is highlighted in the revised version.

Comment 4. It is not justifiable to exclude the NEC-risk subgroup from the between-group comparisons.

Response 4. Thank you for this comment. We agree that the rationale for handling the NEC-risk subgroup should be described. In our study, infants classified as NEC-risk (Bell stage I) represented a clinically heterogeneous group with nonspecific and often transient gastrointestinal symptoms that do not indicate confirmed inflammatory disease and are not equivalent to systemic infection. Including these infants in the primary EONS vs. non-EONS comparisons would increase misclassification bias and obscure SCFA patterns specific to confirmed early-onset sepsis. For this reason, the NEC-risk subgroup was retained for descriptive reporting only, but not incorporated into the between-group statistical analyses. This justification has now been added to the “Methods” section and highlighted in the revised manuscript.

 

Comment 5. Table 1 is a result and should be in the result section.

Response 5. Thank you for this comment. We agree that Table 1 summarises baseline cohort characteristics and therefore belongs in the Results rather than the Methods. Accordingly, Table 1 has been moved to the beginning of the Results section, where the study population is introduced. The revised placement is now highlighted in the manuscript.

 

Comment 6. Table 1 need to include type antibiotic use and timing of first antibiotic use in both groups as antibiotics are often given in preterm babies at birth and may influence the gut microbiome. The timing of the first stool sample on day 3 will have been after onset of early onset sepsis and hence first antibiotic administration in the majority of patients. Stool profiles therefore reflect result of sepsis and result of treatment intervention and not predisposition. The authors need to carefully and precisely report and analyse exact timing of first antibiotic administration and first stool sampling as well as all data on inflammatory markers and their timing after birth.
Response 6. Thank you for this important comment. We have added detailed information on antibiotic exposure to the Results and clarified its interpretive implications in the Discussion. Specifically, we now report that all infants with EONS received empiric amoxicillin–gentamicin therapy within the first 6 hours of life, and the exact timing of antibiotic initiation is explicitly stated. Because TP1 stool samples were collected at approximately 3 days of life, this sampling window necessarily reflects the period after both the onset of systemic infection and the initiation of therapy.

We also clarify that antibiotic exposure was common in the non-EONS group: 27 of 31 newborns received empirical antibiotics starting on the first day of life, and in 10 of these infants treatment was discontinued after 48 hours once infection had been excluded. This information has been added to Table 2 and the Results section.

Importantly, the specific pattern of SCFA alterations in the EONS group—particularly the loss of acetate dominance and the relative enrichment of branched-chain and other non-acetate acids—aligns more closely with metabolic responses to systemic inflammation than with antibiotic exposure alone. These contextual details have been incorporated into the Discussion, where we also acknowledge that infection and empiric treatment overlap in timing and that their individual contributions cannot be fully disentangled within this design. All corresponding revisions are highlighted in the manuscript.

 

Comment 7. There is no sample size calculation presented. The study was underpowered to investigate NEC.

Response 7. Thank you for this comment. As this was an exploratory pilot cohort, no a priori sample size calculation was performed. The study was not powered to analyse NEC-specific outcomes, and the NEC-related findings are therefore descriptive and hypothesis-generating. We now explicitly state this limitation in the Discussion. The lower-than-expected NEC incidence during enrolment further restricted the ability to conduct powered NEC analyses. Corresponding clarifications have been added and highlighted in the revised manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

I commend you on the completion of this pioneering pilot study, which examines the longitudinal fecal short-chain fatty acid (SCFA) trajectories in preterm infants with early-onset neonatal sepsis (EONS). The longitudinal approach is crucial for understanding the metabolic dynamics of the gut in this vulnerable population, and the emphasis on SCFA ratios is particularly appropriate.

I believe that the article, following the revisions detailed below, will be ready for publication. My comments are organized by section, focusing on methodological clarity and the interpretation of results.

With the suggested changes, the article is suitable for publication.

Sincerely,

A Peer Reviewer

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

We sincerely thank you for your careful reading of our manuscript and for the thoughtful and constructive comments provided. Your detailed evaluation has been extremely valuable in improving the clarity, methodological transparency, and interpretability of our work. We carefully addressed each of your points and have revised the manuscript accordingly. All changes are highlighted in the updated version.

Below we provide point-by-point responses to your comments.

Comment 1. The finding that a decrease in acetic acid dominance and an enrichment of non-acetic and branched-chain fatty acids (BCFAs) were evident at TP1 is the core finding and key to the study. In the abstract, ensure you emphasize that this finding is early (TP1) and represents structured alterations.

Response 1. Thank you for this helpful comment. We have revised the Results section of the abstract to emphasise that the SCFA shifts observed at TP1 represent distinct early changes in SCFA composition. We also note that the Conclusions section of the abstract already refers to these findings as “early, structured alterations” in fecal SCFA profiles. The revised wording in the Results now aligns with and reinforces this description. All edits are highlighted in the updated abstract.

Comment 2. The punctuation in the phrase "SCFAs-including acetic, propionic, and butyric acids..." is not optimal. Suggest: Change the dash to parentheses or commas. E.g., "SCFAs (including acetic, propionic, and butyric acids...)".

Response 2. Thank you for this comment. We fully accepted your suggestion and replaced the dash with parentheses. The phrase now reads: “SCFAs (including acetic, propionic, and butyric acids) and related branched-chain fatty acids (BCFAs)”. The revision is highlighted in the manuscript.

 

Comment 3. The explanation for the shift in research focus from NEC to EONS due to the low incidence of NEC is necessary for the transparency of the pilot design.

Response 3. Thank you for these important comments. We have substantially revised the Introduction to make the pilot design and the shift in focus fully transparent. We now explain that the study was originally motivated by the pathophysiology of NEC, but that the incidence of NEC in the targeted gestational window during the enrolment period was unexpectedly low, making adequately powered NEC-focused analyses unfeasible. Consequently, the primary analyses were refocused on infants with and without early-onset neonatal sepsis (EONS), while the original high-resolution, time-structured sampling framework was retained (highlighted in the revised manuscript).

We also rephrased the same paragraph to explicitly state that NEC and EONS share key axes of gut–immune interaction—dysbiosis, impaired epithelial barrier function and early inflammatory signalling—thereby justifying the continued use of the longitudinal SCFA sampling protocol in this EONS-centred analysis. In addition, the subsequent paragraph now provides a detailed biological and clinical rationale for extending SCFA profiling across the entire first postnatal month, with supporting references on early-life microbiome dynamics, antibiotic effects, and longitudinal changes in SCFAs and fecal biomarkers in preterm infants. All corresponding changes are highlighted in the revised Introduction.

 

Comment 4. Inclusion Criteria: Greater clarity is needed regarding which groups were included in the main comparative analysis. Clarity Requirement: On lines 116-119, where you mention infants >32 weeks with suspected NEC, it must be explicitly clarified that these cases were NOT included in the main EONS vs. non-EONS statistical test (as implied later in the manuscript).

Response 4. Thank you for this helpful comment. We have now explicitly clarified in the Methods section that infants with suspected NEC and gestational age > 32 weeks constituted an ancillary descriptive cohort and were not included in the primary EONS vs. non-EONS statistical analyses. This statement has been added directly to the relevant paragraph and is highlighted in the revised manuscript.

 

Comment 5. Time Points (TPs): The five time points (TP1-TP5) are clear as centers (e.g., TP1 3 DoL). For reproducibility, please specify the window range for each TP (e.g., TP1: Days of Life (DoL) 2-4; TP2: DoL 6-8) to account for collection variability.

Response 5. Thank you for this suggestion. We have now added the explicit collection ranges for each time window to improve reproducibility. The effective ranges based on observed data are: TP1 (DoL 1–5), TP2 (DoL 6–11), TP3 (DoL 11–19), TP4 (DoL 17–26), and TP5 (DoL 26–30). This information has been added to the Methods section and highlighted in the revised manuscript.

 

Comment 6. Patient Characteristics (Table 1 Context): Table 1 shows that the EONS group was significantly sicker (lower birth weight, Apgar scores, higher mortality, more respiratory support). This difference in disease severity must be acknowledged in the Discussion as a potential confounding factor. Explain that the use of SCFA ratios mitigates the effect of total stool mass/disease severity, allowing for a clearer focus on metabolic dysbiosis versus overall inflammatory state.

Response 6. Thank you for this important comment. We agree that infants in the EONS group exhibited greater overall illness severity, including lower birth weight, lower Apgar scores, more frequent respiratory support, and higher mortality. We have now added a dedicated statement to the Discussion acknowledging these differences as potential confounding factors. We also clarify that our reliance on ratio-based SCFA metrics provides internally normalized measures that reduce the influence of total stool mass and global disease severity, allowing clearer interpretation of metabolic dysbiosis associated specifically with EONS. The corresponding additions are highlighted in the revised manuscript.

 

Comment 7. Feeding Restriction: Acknowledging that no adjustment was made for the absolute volume of enteral feeding and its impact on MCFAs is an honest recognition of a limitation.

Add a concluding sentence stating that, despite this limitation, the longitudinal data collection provides greater internal consistency and reliability than cross-sectional studies.

Response 7. Thank you for this thoughtful and constructive comment. We agree that the lack of detailed, longitudinal enteral feeding data represents an important limitation, particularly for interpreting medium-chain fatty acids. Although aggregated indicators of early enteral nutrition are now reported in Table 2, these measures do not provide sufficient temporal resolution to adjust SCFA or MCFA levels for substrate availability. As suggested, we have added a concluding sentence noting that, despite this limitation, the longitudinal, repeated-measures structure of the study provides substantially greater internal consistency than cross-sectional designs. The revised text is highlighted in the manuscript.

 

Comment 8. Limitations: Acknowledging the modest cohort size and the lack of corresponding microbiome/immunological profile is correct and serves as a clear call for future research.

Response 8. Thank you for highlighting this point. These limitations are indeed important for contextualising our findings. We have explicitly acknowledged both the modest cohort size and the absence of accompanying microbiome or immunological profiling in the Discussion, noting that these factors constrain mechanistic interpretation and underscore the need for larger, integrated multi-omics studies.

 

Comment 9. Overall Recommendation: This is a very good pilot study that provides the first longitudinal evidence of gut metabolic alterations during EONS. The authors should implement the minor corrections for methodological clarity and punctuation.

Response 9. We sincerely thank the reviewer for the positive overall assessment of our work and for the constructive suggestions provided. We appreciate the acknowledgement of the study’s contribution, and we are grateful for the reviewer’s careful evaluation, which has undoubtedly helped improve the clarity and quality of the manuscript. All revisions have been implemented and are highlighted in the revised version.

Reviewer 3 Report

Comments and Suggestions for Authors

This is a well-structured and well- designed pilot cohort study. Methodology is clearly described and the conclusions are well-presented. 

I suggest minor points for clarification:

1. Sample size may affect statistical results, as it is low.
2. Exposure in antibiotics could be included, as a factor which affects SCFA levels.
3. Feeding methods could also be included in order to highlight potential impact upon SCFA levels

Author Response

Dear Reviewer,

We sincerely thank you for your positive evaluation of our work and for the constructive comments provided. We greatly appreciate your thoughtful suggestions, which have helped us to clarify several important aspects and further improve the manuscript. Below we address each of your points in detail.

 

Comment 1. Sample size may affect statistical results, as it is low.

Response 1. Thank you for this comment. We agree that the modest cohort size limits statistical power. As also addressed in our response to Reviewer 1 and now explicitly stated in the Discussion, this was an exploratory pilot cohort and no a priori sample size calculation was performed. Consequently, the study was not powered to investigate NEC-specific outcomes, and all NEC-related observations should be regarded as descriptive and hypothesis-generating. This clarification has been added to the Limitations section and is highlighted in the revised manuscript.

 

Comment 2. Exposure in antibiotics could be included, as a factor which affects SCFA levels.

Response 2. Thank you for this important comment. We have now added clarifying information on antibiotic exposure to the Results section, specifying the timing of empiric therapy initiation in infants with EONS and noting that TP1 stool samples were collected after both clinical recognition of sepsis and treatment onset. We also clarify that empiric antibiotics were administered to most infants without EONS (27/31), typically initiated on the first day of life and discontinued after 48 hours in 10 cases once infection or pneumonia had been excluded. In the Discussion, we further address the potential influence of antibiotics on early SCFA patterns while explaining why the specific metabolic shifts observed in the EONS group are more consistent with inflammatory responses than with antibiotic exposure alone. Corresponding revisions are highlighted in the manuscript.

 

Comment 3. Feeding methods could also be included in order to highlight potential impact upon SCFA levels.

Response 3. Thank you for this helpful comment. Information on feeding modality (exclusive human milk, donor milk, or mixed enteral feeding) was already collected, and no qualitative differences between groups were observed; therefore, we did not modify Table 1. We agree, however, that not only the type but also the volume and progression of enteral feeding may influence absolute SCFA levels, particularly medium-chain fatty acids. Detailed quantitative feeding-volume data were not consistently available for this pilot cohort, and we now explicitly acknowledge this limitation in the Discussion. We also note that future studies should incorporate standardised nutritional data to better assess the contribution of feeding to SCFA variability.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors now need to conduct a multiple logistic regression analysis entering all significantly different baseline characteristics and outcomes as independent and group allocation as dependent variable to check whether the outcome differences are related to those baseline differences.

 

 

Author Response

Dear Reviewer,

We sincerely thank you for your thorough assessment and for raising the important methodological question regarding the use of multiple logistic regression. We fully agree that careful consideration of potential confounding is essential, and we appreciate the opportunity to clarify the analytic constraints of this pilot dataset. Below we provide an explanation of why multivariable regression is not statistically appropriate or methodologically valid in this context, followed by a description of the sensitivity analysis we conducted to address the underlying concern. Corresponding clarifications have been added to the revised manuscript.

Comment. The authors now need to conduct a multiple logistic regression analysis entering all significantly different baseline characteristics and outcomes as independent and group allocation as dependent variable to check whether the outcome differences are related to those baseline differences.

Response. We fully appreciate the motivation behind the recommendation, and we agree that such modelling is valuable in sufficiently powered analytic datasets. However, in this pilot study, a multiple logistic regression is not feasible for several reasons:

1. Sample size limitations preclude statistically valid multivariable modelling. The EONS group contains 18 infants, which is insufficient to support a stable logistic regression model. Standard methodological recommendations require at least 10–15 observations in the smaller outcome group per predictor variable. With only 18 events, even a model containing two covariates would violate this criterion. Any adjusted regression—including several baseline characteristics—would therefore risk overfitting, coefficient instability, and non-interpretable results.
   
   
2. Several of the “baseline differences” cannot be included as predictors because they are consequences of EONS. For a variable to function as a predictor in a regression model, it must precede the outcome and not be causally downstream from the condition being predicted. Measures such as low Apgar scores, early respiratory deterioration, surfactant treatment, mechanical ventilation, prolonged intensive care, or mortality reflect the clinical course of early-onset sepsis itself rather than independent pre-exposure characteristics. Including such variables would introduce reverse causation and circularity, making the model conceptually invalid regardless of sample size.
   
   
3. The purpose of this pilot study was exploratory metabolic characterisation, not adjusted causal inference. The study was not powered for multivariable modelling, and no a priori sample size calculation was planned for this purpose. Developing an adjusted predictive model falls outside the intended analytic scope of this pilot design.
   
   
4. Sensitivity analysis conducted as an alternative. To address the core question—whether early SCFA differences might simply reflect greater illness severity—we performed a targeted sensitivity analysis. We repeated TP1 comparisons after excluding infants with fatal outcomes or prolonged invasive ventilation. The characteristic metabolic pattern in the EONS group (reduced acetate dominance and enrichment of non-acetate and branched-chain SCFAs) remained unchanged. This suggests that the observed metabolic alterations are unlikely to be driven solely by severity-related factors.
   
   

A brief description of this analysis has been added to the Discussion, and the relevant passages have been revised to reflect this clarification.

We agree that in larger, multicentre cohorts with adequate power, formal adjusted modelling will be essential to disentangle severity-related contributions from EONS-specific metabolic signatures. We now note this explicitly in the Limitations section.

We thank the reviewer again for this constructive comment and for helping strengthen the transparency and clarity of the manuscript.

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed the comment sufficiently.