Round 1

Reviewer 1 Report (New Reviewer)

Comments and Suggestions for Authors Dear Authors, I have studied your manuscript “SLC6A4 Methylation and Kangaroo Mother Care in Preterm Infants: A Longitudinal and Prospective Study” where you explore the role of Kangaroo Mother Care (KMC) on SLC6A4 methylation development and future epigenetic trajectories of preterm infants. The proposed issue is timely and relevant in the current context, particularly since growing concerns about early stage epigenetic influences on epigenetic systems are emerging. The longitudinal study design and recurrent methylation measures represent several strengths. But a number of methodological, analytical and structural aspects, analytical or methodological and presentation problems should be understood or clarified prior to any publication consideration of a manuscript that can be introduced/used. Here are my some minor recommendations I suggest some suggested minor corrections in order from beginning to end, which are the following:   1.0 Title and Abstract.   1.1 The title has redundancy (”Methylation of the SLC6A4 Methylation and gene […]”); it needs to be rewritten for clarity and conciseness.   1.2 In the abstract, you note: “Methylation levels…was stable over time”; grammatically correct by “were stable”.   1.3 The summary in the abstract overgeneralizes the CpG6 result; since the result did not gain significance in the multiple testing corrections, the phrasing should be revised.   2.0 Introduction.   2.1 The background related to SLC6A4 polymorphisms and methylation is appropriate, but a clearer transition could be made connecting genetic variation to epigenetic regulation, to promote conceptual cohesion.   2.2 Human studies on KMC and SLC6A4 methylation do not exist, that claim should be better supported with a systematic rationale (e.g., brief mention of the search strategy or search scope).   2.3 Figure 1 is informative not central; shortening legend, and focusing on relevancy to study hypothesis.   3.0 Methods.   3.1 Some typographical and formatting inconsistencies (e.g., duplicate section numbering: “2.24,” “2.35”) should be corrected for clarity.   3.2 The KMC exposure definition (binary classification) oversimplifies the variability in dose/intensity, this is justified but should be noted earlier in the Methods.   3.3 A strength of the DAG-based confounder selection is that, it is unclear and written in grammatical error. This sentence is not a clean, concise description of the DAG approach and could be improved for both clarity and methodological veracity.   3.4 For blood samples, storage conditions (up to 72h and 2–8°C) should include a justification or reference confirming DNA methylation stability at these storage conditions.   3.5 No a priori power calculation was carried out; that is, while exploratory, this paper should have discussed quantified and detectable effect sizes in some post hoc manner to promote the methodological clarity.   4.0 Results.   4.1 There are many redundancies and inconsistencies (e.g. repeated section titles) (e.g. 3.1 Characteristics of the Study Population).   4.2 Several numbering errors can be found in tables and figures (e.g., “Figure 43”, “Table 43”, “Figure 54") which should then be worked systematically.   4.3 Standardization in the report of CpG methylation ranges (for example, disagreement between text-reported and supplementary material) will be needed.   4.4 Reporting of statistical results should include the uniformity in the assignment of precisely computed p-values and confidence intervals.   4.5 More explicitly addressing group imbalance (51 vs 24) when interpreting null results are required in the interpretation.   5.0 Statistical Analysis.   5.1 Justification for FDR < 10% is fine for exploratory work, however more discussion is needed for false positive analysis.   5.2 The computation and interpretation of effect sizes (Cohen’s d) are appropriate but their clinical significance should be provided with some context.   5.3 “No correction for multiple testing was carried out” in some analyses might be qualified as “no correction for multiple testing was carried out” to avoid confounding with changes in previous FDR’s.   6.0 Discussion.   6.1 General discussion is thorough, with verbose sections (e.g., long explanation of animal studies), the analysis could be edited.   6.2 Multiple or partially edited sentences appear in several paragraphs showing incomplete revision (i.e., the explanations of NR3C1 studies overlapping).   6.3 The interpretation of the CpG6 finding should be purely exploratory and not suggestive of biological relevance without functional validation.   6.4 The possibility of reduced group differences in comparison to “humanized care” is interesting but speculative; this idea should be clearly articulated.   6.5 Pain/stress quantification is a significant constraint and should be more well addressed.   7.0 Limitations.   7.1 Study limitations are described (in part) and should be summarised in a full sub-section.   8.0 Language & Style.   8.1 The manuscript requires careful editing for its English language as we have seen several errors in grammar, inconsistency, and typographical errors through it all.   8.2 Certain sentences are unnecessary because they're long as well as repetitive, which must be cut down to just make them easier for you to read.   8.3 Formatting problems (inconsistent fonts, spacing and alignment) are present and must be fixed as described in the journal instructions.   9.0 Figures and Related Information.   9.1 Make sure everything else on the supplementary table is referenced properly and should be properly identified and numbered clearly.   Conclusion. Using an appropriate longitudinal design, the manuscript deals with an otherwise relevant, but yet an under-researched issue, the well-established question. A big editorial revision, clarification on the methodology and consistent reporting are, however, necessary. It may provide scientific benefits, but clearer reporting and cautious interpretation of findings are required.  Kind regards

Author Response

Response to Reviewer 1

General comment: The reviewer noted a timely and relevant topic, praised the longitudinal design, but identified methodological, analytical, and structural aspects requiring clarification.

Response: We thank Reviewer 1 for the thorough reading. We have carefully addressed each point below.

1.0 Title and Abstract

1.1 – Title redundancy

Reviewer comment: The title has redundancy ("Methylation of the SLC6A4 Methylation and gene […]"); it needs to be rewritten for clarity and conciseness.

Response: We were surprised by the reviewer's observation regarding the redundancy in the manuscript title, which was actually "SLC6A4 gene methylation and Kangaroo Mother Care in Premature Infants: A Longitudinal and Prospective Study". We believe there may have been a formatting error during the submission process. Anyway, the title has been revised and changed to ensure the absence of redundancy and to meet the second reviewer’s request for greater clarity, as follows: "SLC6A4 gene methylation in Premature Infants undergoing Kangaroo Mother Care: A Prospective Longitudinal study".

1.2 – Subject-verb agreement in abstract

Reviewer comment: "Methylation levels…was stable over time" — should be "were stable."

Response: We changed it to "Methylation levels…remained stable over time."

1.3 – Overgeneralization of CpG6 result in abstract

Reviewer comment: The abstract overgeneralizes the CpG6 result; since it did not survive correction for multiple testing, the phrasing should be revised.

Response: We have revised the Abstract's conclusion to reflect that no robust associations were found and that the nominal difference at CpG6 is, at most, hypothesis-generating rather than definitive. The abstract conclusion now reads:

“KMC was not associated with major longitudinal changes in SLC6A4 methylation during the neonatal period. The nominal difference at CpG6 should be interpreted as exploratory and warrants further investigation. Larger, multicenter studies with long-term follow-up are needed to clarify the epigenetic mechanisms linking early caregiving experiences with stress regulation and neurodevelopmental outcomes in preterm infants.”

2.0 Introduction

2.1 – Transition from genetic to epigenetic regulation

Reviewer comment: A clearer transition connecting genetic variation to epigenetic regulation would improve conceptual cohesion.

Response: We agree with the reviewer that the conceptual link between genetic variation and epigenetic regulation needed further clarification. We have revised the introductory paragraph to include a more explicit transition, as follows:

“Beyond genetic variation, the promoter region of SLC6A4 is also subject to epigenetic regulation via DNA methylation, which may represent an additional layer of control that can be modulated by early environmental exposures."

2.2 – Absence of human KMC/SLC6A4 studies

Reviewer comment: The claim that no human studies exist on KMC and SLC6A4 methylation should be better supported with a brief rationale or search scope.

Response: We have altered the sentence regarding the absence of studies in humans and added a brief statement documenting the bibliographic research that supports this claim, as follows:

"A systematic search of PubMed and LILACS using the terms 'Kangaroo Mother Care', 'skin-to-skin contact', 'SLC6A4', and 'methylation' did not retrieve any human studies that specifically examined this association."

2.3 – Figure 1 legend

Reviewer comment: Figure 1 legend should be shortened and focused on its relevance to the study hypothesis.

Response: We thank the reviewer for this suggestion. The legend was abbreviated to focus on the serotonergic synapse and the role of SLC6A4 as the reuptake transporter central to this study. Now it reads:

"Schematic of the serotonergic synapse.  The serotonin transporter (5-HTT), encoded by the SLC6A4 gene, is the primary mechanism for serotonin (5-HT) reuptake from the synaptic cleft, regulating the duration and intensity of the signal, and is the focus of this study”.

3.0 Methods

3.1 – Section numbering inconsistencies

Reviewer comment: Duplicate section numbering ("2.24", "2.35") should be corrected.

Response: Again, we believe there may have been a formatting error during the submission process. In any case, all section numbering has been reviewed, and now reads:

2.1. Study design and study population, and Ethics

2.2. The Kangaroo Mother Care Method

2.3. Ethics

2.4. Data Collection and Clinical Variables

2.5. Procedures for Gene Methylation Analysis

2.6. Statistical Analysis

3.2 – Binary classification of KMC

Reviewer comment: The binary KMC classification justification should appear earlier in the Methods.

Response: The justification has been repositioned to the beginning of the statistical section, as suggested by the reviewer.

3.3 – DAG description clarity

Reviewer comment: The DAG-based confounder selection description is unclear and has a grammatical error.

Response: We apologize for the grammatical errors and the lack of clarity regarding the DAG-based covariate selection. We have revised the description of the DAG-based covariate selection to improve clarity and correct grammatical errors. We now explicitly distinguish between the full set of variables used to construct the causal framework and the specific minimal adjustment set selected for the final analysis. We included a paragraph clarifying that among the four mathematically equivalent sets identified by the DAG, a specific set was selected for our models to ensure clinical relevance and data robustness. This paragraph was previously in the ‘Results’ section, but we have now moved it to the “Materials and Methods’ section, which now reads:

"To identify potential confounding variables for the causal association between Kangaroo Mother Care and SLC6A4 methylation changes over time, we developed a Directed Acyclic Graph (DAG). The DAG framework incorporated maternal and gestational factors (education, marital status, maternal diseases, gestational complications, antenatal corticosteroids, and prenatal care) as well as neonatal variables (gestational age, birth weight, sex, Apgar at 5 minutes, CRIB (Clinical Risk Index for Babies), and the highest NTISS (Neonatal Therapeutic Intervention Scoring System), early and late neonatal complications) (Figure 3).

Among the four minimal sufficient adjustment sets identified by the DAG, we selected the following set for inclusion in our multivariable models: sex, gestational age, early neonatal complications, late neonatal complications, and the highest NTISS score. This approach ensures that the causal effect is estimated while avoiding overadjustment and bias."

3.4 – Blood sample storage justification

Reviewer comment: Storage conditions (up to 72h at 2–8°C) should include a reference confirming DNA methylation stability.

Response: A supporting reference has been added confirming that bisulfite pyrosequencing-based methylation profiles remain stable under these storage conditions (Li Y et al. Stability of global methylation profiles of whole blood and extracted DNA under different storage durations and conditions. Epigenomics. 2018 Jun;10(6):797-811. doi: 10.2217/epi-2018-0025). The authors evaluated global methylation profiles using Reduced Representation Bisulfite Sequencing (RRBS) in whole blood and extracted DNA samples subjected to different storage conditions, including refrigeration at 4 °C, freezing, and prolonged storage for up to 20 years. The authors observed a high correlation among methylation profiles obtained under the different conditions analyzed, with no statistically significant differences in the overall similarity of profiles between recently processed samples and those stored for extended periods. In addition, the study demonstrated that any storage-related alterations occurred consistently across samples, preserving the validity of comparative methylation analyses. Therefore, the available evidence suggests that refrigerated storage of samples does not substantially compromise the overall integrity of methylation profiles for comparative epigenetic analyses.

 

The text now reads:

“All samples were collected by a NICU neonatologist in EDTA tubes, stored at 2-8°C in the maternity ward for up to 72 hours, a condition previously shown to preserve relative stability of global DNA methylation profiles [17]. They were then transported to the epigenetics laboratory, where they were aliquoted and stored at -80°C until processing”.

 

We also added an excerpt in the Strengths and Limitations subsection of the Discussion section (4.6), acknowledging this issue as a limitation, as follows:

 

“For example, it can be questioned whether the initial storage conditions for the samples (at 2-8°C for up to 72 hours) were adequate. Although some studies recommend shorter intervals between sample collection and processing for methylation analyses, this issue is still under debate. Furthermore, the methylation levels in our study remained stable over time, suggesting no systematic temporal drift or increased variability indicative of instability in methylation profiles associated with storage conditions. We believe that the pre-analytical conditions used did not introduce a relevant bias into the analyses.”

3.5 – Power calculation

Reviewer comment: No a priori power calculation was carried out; some post hoc discussion of detectable effect sizes should be included.

Response: We acknowledge this limitation. No prior data on KMC-specific SLC6A4 methylation existed to support a priori calculation. Rather than presenting a post hoc power calculation — which can be misleading in observational contexts — we explicitly acknowledged in the Limitations that the study was likely underpowered to detect small effect sizes, and contextualized the sample size against comparable published epigenetic studies in neonatal care (e.g., Provenzi et al. 2015, n=56; Montirosso et al. 2016, n=48). Results are framed as hypothesis-generating throughout. In any case, we have reinforced the interpretation of the results as exploratory and hypothesis-generating, as follows:

“Thus, results need to be interpreted as exploratory and hypothesis-generating. While a formal a priori power calculation was not feasible due the lack of prior data on the association between SLC6A4 methylation and KMC, our sample size is comparable to other published epigenetic studies in the field of neonatal intensive care.”

4.0 Results

4.1 – Redundant section titles

Reviewer comment: Repeated section titles (e.g., "3.1 Characteristics of the Study Population") should be removed.

Response: We apologize for this. Again, we believe that some formatting issues may have occurred during submission. Anyway, we reviewed the manuscript to ensure there are no duplicate section titles.

4.2 – Numbering errors in figures and tables

Reviewer comment: Numbering errors such as "Figure 43", "Table 43", "Figure 54" must be corrected.

Response: Again, we believe that some formatting issues may have occurred during submission. In any case, we reviewed the manuscript to ensure there are no numbering errors.

4.3 – CpG methylation range standardization

Reviewer comment: Disagreement between text-reported and supplementary methylation ranges must be resolved.

Response: We thank the reviewer for identifying this discrepancy. We have corrected the methylation ranges in the text to precisely match the median values presented in Table S3. Specifically, the lower bound for CpGs 12 and 13 was corrected from 7.16% to 7.19%. We also clarified that these ranges refer to the median methylation percentages across groups and time points. Now it reads:

“Descriptive statistics are available in the supplementary material (Table S3). Median methylation percentages across all time points ranged from 0.78% to 2.78% for CpGs 1-11. Sites 12 and 13, showed consistent higher methylation levels, with medians ranging from 7.19% to 10.76%.” 

4.4 – Statistical reporting uniformity

Reviewer comment: Uniform reporting of p-values and confidence intervals is needed throughout.

Response: We have standardized the reporting of all statistical results throughout the manuscript. Specifically, we have:

1. Ensured that all p-values are reported with three decimal places.
2. Consistently italicized the 'p' symbol.
3. Provided 95% Confidence Intervals (95% CI) alongside p-values for all primary outcomes in the text and tables, as requested.

4.5 – Group imbalance and null result interpretation

Reviewer comment: The group imbalance (51 vs 24) should be addressed explicitly when interpreting null results.

Response: Explicit discussion of the group imbalance has been added to the Limitations section (4.6), noting its potential effect on statistical power, as follows:

“The relatively small sample size (due to logistical and financial constraints), the low absolute methylation levels at most CpG sites, and the imbalance between groups (51 KMC vs. 24 non-KMC) likely limited the power to detect small effect sizes. It should be considered when interpreting null findings.”

5.0 Statistical Analysis

5.1 – FDR threshold justification

Reviewer comment: More discussion is needed on the FDR 10% threshold and its implications for false positives.

Response: We acknowledge that a 10% FDR threshold carries a higher risk of Type I errors (false positives) compared to the traditional 5% level. This threshold was chosen a priori to avoid an overly conservative approach that might fail to detect subtle biological signals (Type II error) in this vulnerable and clinically complex population. We have expanded the justification in the Methods and its implications in the limitations to emphasize that since no CpG site reached statistical significance even at this 10% threshold, the risk of reporting a false-positive finding in this study is effectively mitigated.

In the Materials and Methods (subsection 2.6):

“Thus, because of the exploratory nature of this study, with a limited sample size and unbalanced groups, a false discovery rate [FDR] of 10% (q < 0.10) was allowed to balance the risk of Type I errors (false positives) against the risk of Type II errors (missing potentially meaningful biological signals in a vulnerable, hard-to-reach population).  This threshold implies that among the results identified as significant, 10% are expected to be false positives. This level is often considered acceptable in discovery-phase studies where the goal is to identify candidates for further validation rather than to provide definitive evidence of association.”

 

In the limitations of the Discussion section (subsection 4.6):

“However, this study was designed as an exploratory analysis of KMC's effect on SLC6A4, which further justifies using a slightly higher FDR threshold to avoid discarding relevant trends. On the other hand, even with this more lenient threshold of 10%, no CpG site reached statistical significance after correction. This suggests that the observed nominal differences are either very small in magnitude or highly variable across individuals, reinforcing the need for caution to avoid false positive interpretations. Thus, results need to be interpreted as exploratory and hypothesis-generating.”

5.2 – Cohen's d clinical significance

Reviewer comment: Clinical significance of Cohen's d should be contextualized.

Response: A brief contextualizing sentence has been added to the Discussion, as follows:

“In addition, this study found no robust differences in longitudinal changes in SLC6A4 gene methylation between those exposed and those not exposed to Kangaroo Mother Care. The predominantly negligible effect sizes (Cohen d. < 0.10) observed across CpG sites are consistent with subtle epigenetic signals typically reported in the early adversity literature and reinforce the exploratory nature of the present findings.”

5.3 – Multiple testing clarification for longitudinal models

Reviewer comment: The statement "no correction for multiple testing was carried out" in longitudinal analyses should be qualified to avoid confusion with the FDR correction applied elsewhere.

Response: To avoid confusion, we made the following changes:

In the Materials and Methods section

“In assessing methylation differences between the two groups at each of the 13 CpG sites, the Benjamini-Hochberg method was used to correct for multiple testing, when nominal differences were statistically significant.” 

In the Results section, subsection 3.3 (Longitudinal Methylation Trajectories):

“Since no p-value reached statistical significance in the longitudinal mixed-effect models, correction for multiple testing was not applied to these analyses.”

6.0 Discussion

6.1 – Verbose sections

Reviewer comment: Verbose sections, especially the animal studies explanation, should be condensed.

Response: The animal studies paragraph has been substantially condensed. Now it reads:

“Human epigenetic research on prematurity is based on the premise that early experiences, both adverse and protective, can be “biologically embedded” through DNA methylation, thereby influencing neurodevelopment [19]. Animal studies have provided a foundation for this, demonstrating that the offspring of low-care mothers exhibit higher methylation and greater stress reactivity, which can be reversed by cross-fostering [10] or pharmacological agents [10, 20]. “

6.2 – Incomplete revision traces

Reviewer comment: Partially edited sentences and overlapping NR3C1 explanations indicate incomplete revision.

Response: Once again, we believe there was a formatting error during the submission process. In any case, we reviewed the entire discussion section to ensure there were no partially edited phrases or overlapping explanations regarding NR3C1.

6.3 – CpG6 interpretive framing

Reviewer comment: The CpG6 finding should be interpreted as purely exploratory, without implying biological relevance absent functional validation.

Response: All mentions of CpG6 in the Discussion now explicitly state that the finding is exploratory and statistically non-robust. We altered the second paragraph of subsection 4.2 in the Discussion section and cut some words in the Conclusions, as follows:

Discussion

“The subtle difference in methylation at CpG site 6 at hospital discharge only in the non-KMC group was not robust after multiple testing correction. Although this finding should be interpreted solely as exploratory, the direction of the finding is consistent with the idea that KMC might act as a buffer against the hypermethylation typically induced by NICU stress.”

Conclusions

“However, these results should be interpreted with caution. They may be due to a genuine lack of effect, a follow-up period that was too short to detect epigenetic changes, or insufficient statistical power to detect subtle effects. On the other hand, the nominal difference in CpG6 methylation at hospital discharge and in CpG 13 methylation trajectories —though isolated and exploratory—warrants further investigation. These findings suggest a possible targeted influence of KMC on SLC6A4 methylation and may represent an early signal of the potential regulatory effects of nurturing care on the serotonergic system. However, the statistical strength of these associations is exploratory, and we frame our results as a foundational step toward future large-scale epigenetic trials in neonatal care.”

6.4 – "Humanized care" speculation

Reviewer comment: The possibility of reduced group differences due to the humanized care environment should be clearly identified as speculative.

Response: We have revised the discussion to clearly identify this possibility as speculative. Now it reads:

“On the other hand, contrary to our expectations, we found no difference between the KMC and Non-KMC groups in any other CpG. One possible, though speculative, explanation is that the institutional humanized care environment may have influenced the results. Since this institution is a national referral center for KMC, we hypothesize that even premature infants whose parents were unable to engage in the method for various reasons received more skin-to-skin contact and gentler handling from the healthcare team compared to conventional care. Although not directly measured, the degree of environmental contrast between the KMC and non-KMC groups may have been insufficient to produce molecular-level changes.”

6.5 – Pain/stress quantification

Reviewer comment: The absence of pain/stress quantification is a significant constraint and should be more thoroughly addressed.

Response: We have expanded the discussion of this constraint, as follows:

“In the present study, we did not quantify the pain level of preterm infants during their NICU stay, which prevents a direct correlation between stress and SLC6A4 methylation. The subtle difference at CpG 6 at hospital discharge was not robust after multiple testing correction. Although, this finding should be interpreted solely as exploratory, the direction is consistent with the idea that KMC might act as a buffer against the hypermethylation typically induced by the stress of the NICU. However, it is not possible to confirm whether a cumulative reduction in perceived stress mediated this.”

7.0 Limitations

7.1 – Dedicated limitations subsection

Reviewer comment: Limitations should be summarized in a full subsection.

Response: A dedicated Limitations subsection (4.6) has been created, consolidating all limitations.

8.0 Language and Style

8.1 – English language editing

Reviewer comment: Several errors in grammar, inconsistency, and typography throughout.

Response: We apologize for the errors observed. We would like to emphasize that we believe a formatting issue occurred during the submission process. In any case, the manuscript has been thoroughly reviewed, and we hope to have resolved all the errors and inconsistencies pointed out.

8.2 – Long and repetitive sentences

Reviewer comment: Unnecessarily long and repetitive sentences must be simplified.

Response: All unduly complex or repetitive sentences have been revised throughout.

8.3 – Formatting inconsistencies

Reviewer comment: Inconsistent fonts, spacing, and alignment must be corrected per journal instructions.

Response: The manuscript has been reformatted to comply with Biomedicines author guidelines.

9.0 Figures and Related Information

9.1 – Supplementary table referencing

Reviewer comment: Supplementary tables should be properly referenced and clearly numbered.

Response: All supplementary tables have been renumbered (S1–S3) and all in-text references corrected.

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

The authors evaluate the association between Kangaroo Mother Care (KMC) and SLC6A4 methylation dynamics in preterm infants using a prospective longitudinal design.

Here’s some comments for improvement:

1. Title: The title is unclear and grammatically incorrect. Suggest revision for clarity.
2. Abstract: The conclusion is overstated as giving protective effect despite non-significant findings. Suggest revision.
3. Methods:

• Sample size calculation is not done. Having said that, this study is likely underpowered to detect subtle methylation changes, with n=75.
• There is no details on how KMC implementation in the NICU. Suggest to add in details.
• Benjamini-Hochberg was used as one of the statistically analyses, but need justification of why setting FDR threshold at 10%, but not standard 5%?

Comments on the Quality of English Language

1. There is obvious weakness in the expression of English language seen throughout the article. Suggest professional English editing service.

Author Response

Response to Reviewer 2

General comment: The reviewer raised concerns regarding the title, the overstated abstract conclusion, the absence of sample size calculation, the lack of KMC implementation details, and the justification for the FDR threshold.

Response: We appreciate the reviewer's focused feedback. Each point is addressed below.

Title

Reviewer comment: The title is unclear and grammatically incorrect. Suggest revision for clarity.

Response: We apologize for the errors. We believe there may have been a formatting error during the submission process. In any case, the title has been revised and changed to ensure the absence of redundancy and to meet the second reviewer’s request for greater clarity, as follows:

"SLC6A4 gene methylation in Premature Infants undergoing Kangaroo Mother Care: A Prospective Longitudinal study".

Abstract – overstated conclusion

Reviewer comment: The conclusion implies a protective effect despite non-significant findings. Suggest revision.

Response: The abstract conclusion has been revised to accurately reflect the null primary findings and to qualify the CpG6 signal as exploratory. It now reads:

“KMC was not associated with major longitudinal changes in SLC6A4 methylation during the neonatal period. The nominal difference at CpG6 should be interpreted as exploratory and warrants further investigation. Larger, multicenter studies with long-term follow-up are needed to clarify the epigenetic mechanisms linking early caregiving experiences with stress regulation and neurodevelopmental outcomes in preterm infants.”

Methods – sample size calculation

Reviewer comment: No sample size calculation was performed. The study is likely underpowered (n=75) to detect subtle methylation changes.

Response: We acknowledge this limitation. This is a convenience sample from a single referral center over one year, within an observational and exploratory study design for which no prior effect size data were available. We explicitly acknowledge in the Limitations that the study may have been underpowered to detect small effect sizes. Nonetheless, the sample size is comparable to published epigenetic studies in the NICU setting (e.g., Provenzi et al. 2015, n=56; Montirosso et al. 2016, n=48), and results are framed as hypothesis-generating, as follows:

“Thus, results need to be interpreted as exploratory and hypothesis-generating. While a formal a priori power calculation was not feasible due to the lack of prior data on the association between SLC6A4 methylation and KMC, our sample size is comparable to that of other published epigenetic studies in the field of neonatal intensive care.”

Methods – KMC implementation details

Reviewer comment: There are no details on how KMC was implemented in the NICU. Suggest adding details.

Response: As mentioned in the Method’s subsection 2.2, the maternity hospital in the study has been a reference center for KMC since 1999. Thus, the method is well established and routinely applied at this institution for a long time. No implementation process was specifically performed for the present study.  The Results subsection 3.2 (KMC implementation results) and Table 3 already present data on the start day, progression through stages, and daily hours. We have expanded this subsection to include who performed it, as follows:

In the Materials and Methods subsection 2.2:

“As a national reference center for KMC since 1999, this maternity ward routinely applies the method, and no specific implementation process was conducted for this study.”

In the Results subsection 3.2:

“Kangaroo Mother Care (KMC) was primarily performed by mothers in both stages, with occasional participation by fathers or other caregivers during stage 1.”

Methods – FDR threshold justification

Reviewer comment: Why was FDR set at 10% rather than the standard 5%? Justification needed.

Response: We acknowledge that a 10% FDR threshold carries a higher risk of Type I errors (false positives) compared to the traditional 5% level. The 10% threshold was chosen to balance the risks of Type I and Type II errors in an exploratory study with a small, hard-to-recruit population and no prior effect size estimates. A 5% threshold would increase the risk of missing biologically meaningful signals. All findings are framed accordingly as exploratory. We have expanded the justification in the Methods subsection 2.6, and its implications in the limitations of the Discussion section (subsection 4.6), to emphasize that since no CpG site reached statistical significance even at this 10% threshold, the risk of reporting a false-positive finding in this study is effectively mitigated.

In the Materials and Methods:

“Thus, because of the exploratory nature of this study, with a limited sample size and unbalanced groups, a false discovery rate [FDR] of 10% (q < 0.10) was allowed to balance the risk of Type I errors (false positives) against the risk of Type II errors (missing potentially meaningful biological signals in a vulnerable, hard-to-reach population).  This threshold implies that among the results identified as significant, 10% are expected to be false positives. This level is often considered acceptable in discovery-phase studies where the goal is to identify candidates for further validation rather than to provide definitive evidence of association.”

 

In the limitations of the Discussion section:

“However, this study was designed as an exploratory analysis of KMC's effect on SLC6A4, which further justifies the use of a slightly higher FDR threshold to avoid discarding relevant trends. On the other hand, even with this more lenient threshold of 10%, no CpG site reached statistical significance after correction. This suggests that the observed nominal differences are either very small in magnitude or highly variable across individuals, reinforcing the need for caution to avoid false-positive interpretations. Thus, results need to be interpreted as exploratory and hypothesis-generating.”

English language quality

Reviewer comment: Obvious weakness in English language expression throughout. Suggest professional editing.

Response: The manuscript has been thoroughly revised for language throughout.

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

This study addresses an important question by exploring the relationship between Kangaroo Mother Care (KMC) and the epigenetic regulation of stress-related genes, specifically SLC6A4. The longitudinal assessment of DNA methylation across multiple time points (birth, NICU discharge, and hospital discharge) is a notable strength and contributes to the growing literature on early-life epigenetic programming in preterm infants. However, several methodological and reporting limitations were present when interpreting the findings.

First, although the study includes preterm neonates (25–34 weeks gestation, birth weight <1800 g), the comparability of the KMC (n=51) and non-KMC (n=24) groups is questionable. Significant differences were observed in key clinical variables such as parenteral nutrition rates and both NICU and total hospital length of stay. These differences suggest that the two groups may not be clinically equivalent at baseline or during hospitalization. Moreover, since biological samples (D2 and D3) were collected at NICU dischange and hospital discharge, the differing lengths of stay imply that the timing of sample collection was not standardized across groups, potentially introducing bias in methylation comparisons.

Second, the inclusion of multiple twin pairs (n=8) in the KMC group raises concerns regarding statistical independence. Without appropriate adjustment (e.g., clustering or mixed models), the presence of correlated observations may bias the results and inflate statistical significance.

Third, important caregiving and nutritional variables are insufficiently described. It is unclear whether KMC was performed by mothers, fathers, or other caregivers, which is relevant given the potential psychosocial and biological implications. Additionally, there is no information on breastfeeding practices, such as timing of first breast milk exposure, feeding type (exclusive breastfeeding vs. mixed feeding), or duration. Given the known influence of human milk on epigenetic regulation and neonatal outcomes, this represents a major unaddressed confounder.

Finally, the study appears to build upon prior work by Brasil AA et al. (2025), which examined SLC6A4 methylation longitudinally in preterm and term infants.

In summary, while the study provides interesting preliminary insights into the potential epigenetic effects of KMC, the presence of baseline imbalances, lack of control for key confounders, unclear analytical strategies, and potential clustering effects limit the strength of the conclusions. Future studies with better group comparability, standardized sampling time points, and multivariable analytical approaches are needed to more definitively elucidate the relationship between KMC and epigenetic regulation in preterm infants.

Author Response

Response to Reviewer 3

General comment: The reviewer acknowledged the study's strengths (longitudinal design, multiple time points, relevant research question) but raised concerns regarding group comparability, statistical independence of twin pairs, insufficient reporting of caregiving variables, and the relationship to a previously published companion study.

Response: We thank Reviewer 3 for the insightful and rigorous critique. We address each point in detail below.

Comment 1 – Group comparability and timing of sample collection

Reviewer comment: The KMC and non-KMC groups differed significantly in parenteral nutrition rates and length of stay. Moreover, since D2 and D3 samples were collected at discharge, differing lengths of stay imply non-standardized sampling timing, potentially introducing bias in methylation comparisons.

Response: We thank the reviewer for raising this important point. Regarding group differences: the observed imbalances in parenteral nutrition and length of stay between groups were expected in an observational convenience sample at a KMC referral center, and likely reflect, at least in part, the clinical effects of KMC itself rather than baseline confounding alone. Both variables were considered in the causal framework (DAG) used for confounder selection. Regarding sampling timing: we fully acknowledge that using NICU and hospital discharge as collection time points — rather than fixed postnatal ages — introduces variability in the age at sample collection between groups. This is an inherent limitation of the study design, as acknowledged in the Limitations section. However, discharge represents a point of achieved physiological stability, a common benchmark in neonatal research that reflects the culmination of the neonatal clinical experience. We expanded the text regarding this limitation, as follows:

 

“Another source of confounding was the use of NICU and hospital discharge as sample collection time points, rather than fixed postnatal ages, resulting in non-standardized timing across groups — particularly given the significantly shorter length of stay in the KMC group. This may have introduced age-related confounding and should be considered when interpreting results."

Comment 2 – Twin pairs and statistical independence

Reviewer comment: The inclusion of 8 twin pairs in the KMC group, without appropriate adjustment for clustering, may bias results and inflate statistical significance.

Response: We thank the reviewer for this insightful observation regarding the twin pairs. We acknowledge that the inclusion of eight twin infants (four pairs) introduces a level of correlation (family-level clustering) that was not explicitly accounted for in our initial linear mixed-effects models, which included only a random intercept at the individual level. However, we believe this does not compromise the study’s conclusions for two main reasons:

1. Direction of Bias: Clustering typically reduces the effective sample size and, if ignored, can artificially deflate p-values. Given that our primary longitudinal findings were non-significant (null results), accounting for family-level clustering would likely result in even larger p-values, further supporting our conclusion that no robust methylation changes were detected.
2. Proportion of Sample: The twin infants represent a small fraction of the total cohort (10.7%; 8/75).

We have added this point to the Limitations section, acknowledging that future studies with larger proportions of multiple births should employ nested random effects to account for family-level correlation fully.

 

“In addition, the inclusion of eight twin infants (four pairs), all in the KMC group, is a limitation as observations within families are correlated. While our linear mixed-effects models included a random intercept per individual, they did not account for family-level clustering. However, since the study’s primary results were non-significant, it is unlikely that accounting for this clustering would have altered the overall conclusions, as adjusting for correlated data typically increases p-values. Nonetheless, this source of bias should be considered in future studies.”

Comment 3 – KMC caregiver identity and breastfeeding

Reviewer comment: It is unclear whether KMC was performed by mothers, fathers, or others. There is also no information on breastfeeding practices (timing, type, duration), which is a known epigenetic confounder.

Response: Regarding caregiver identity: KMC was primarily performed by mothers, in accordance with the Brazilian KMC protocol. In some cases, fathers and other primary caregivers participated during stage 1. A clarifying sentence has been added in the Results subsection 3.2.

In the Results subsection 3.2:

“Kangaroo Mother Care (KMC) was primarily performed by mothers in both stages, with occasional participation by fathers or other caregivers during stage 1.”

Regarding breastfeeding: We thank the reviewer for this relevant point. We acknowledge that although all infants in our cohort received human milk (maternal and/or pasteurized donor milk), the degree of exclusivity and the use of supplemental formula varied across groups. In our cohort, the KMC group had a higher proportion of exclusive breastfeeding at hospital discharge compared to the non-KMC group (as reported in our previous clinical study*). We did not include breastfeeding type as a covariate in the current SLC6A4 methylation models for two reasons:

1. Detailed longitudinal data on the daily volume and duration of breastfeeding were not available;
2. Breastfeeding is intrinsically linked to the KMC intervention itself, acting as a potential mediator. Adjusting for a mediator can sometimes lead to 'overadjustment bias' when the goal is to assess the total effect of the KMC intervention.

We have updated the Limitations section to acknowledge that the lack of granular breastfeeding data prevents us from isolating its specific epigenetic contribution, as follows:

"A further issue was the lack of detailed quantification of breastfeeding practices. Although human milk was provided to the entire cohort, the KMC group achieved higher rates of exclusivity at discharge. Since breastfeeding is a known epigenetic modulator, it may mediate the effects of KMC. Without precise data on the breastfeeding 'dose' over time, we could not distinguish its independent effect from the overall effect of KMC."

 

* Campanha PPA, de Magalhães-Barbosa MC, Prata-Barbosa A, Rodrigues-Santos G, da Cunha AJLA. Exclusive breastfeeding and length of hospital stay in prematureinfants at a Brazilian reference center for kangaroo mother care. J Pediatr (RioJ). 2024 Jul-Aug;100(4):392-398. doi:10.1016/j.jped.2024.01.004. Epub 2024 Mar 21. PMID: 38522479; PMCID: PMC11331221.

Comment 4 – Relationship to prior publication (Brasil et al., 2025)

Reviewer comment: The study appears to build upon prior work by Brasil et al. (2025), which examined SLC6A4 methylation longitudinally in preterm and term infants.

Response: The previous study (reference 35) by Brasil et al. (2025) [ was conducted by authors from the same working group as the present study, focusing on epigenetics and neurodevelopment. However, that study examined SLC6A4 methylation in a distinct cohort of infants from a different maternity hospital, comparing preterm and full-term infants in the first three months of life, without considering KMC exposure. The present study focuses on a different cohort of preterm infants, with or without KMC exposure, and examines SLC6A4 methylation from birth through hospital discharge. Both studies use the same pyrosequencing platform and CpG region, allowing for comparisons. The distinction between the two cohorts is stated in section 4.2 of the Discussion and has been further clarified in the revised manuscript. Now it reads:

“Furthermore, in the present study, we did not compare SLC6A4 methylation at birth between preterm infants and full-term newborns. However, in a recently published study [35], we made this comparison in a distinct cohort of newborns from another maternity hospital.”

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. Exposure measurement (“dose” of KMC) is insufficient
KMC is treated as a binary exposure (completed two in-hospital phases) without quantifying daily/cumulative skin-to-skin hours, frequency, or maternal presence. This limits dose–response inference and comparability.
Please: (i) add quantitative KMC metrics (e.g., cumulative hours, average hours/day, number of days) and run dose–response or sensitivity analyses; or (ii) explicitly discuss how the lack of dose may dilute effects.

2. Baseline imbalance and potential confounding
The KMC group showed higher maternal education, more stable marital status, more ≥6 prenatal visits, and—on the infant side—less parenteral nutrition and shorter NICU/total length of stay. These socioeconomic and care-intensity differences may influence epigenetic marks.
Please: incorporate key sociodemographic variables (education, marital status, prenatal care) and care-intensity variables (parenteral nutrition, NICU days/LOS) into your causal framework and adjust for them in the mixed models or provide sensitivity analyses.

3. Sample size, multiple testing, and interpretation
With n=75 (KMC 51 vs non-KMC 24) and 13 CpGs across multiple time-points, power is limited. While you used BH correction at FDR 10% (exploratory), the rationale and the trade-off between type I/II errors need clearer justification.
Please: report effect sizes with 95% CIs for key contrasts (not just p/q values); clarify any a priori/post hoc power considerations; explicitly justify q<0.10 and what it implies for inference strength.
4. Tissue-source heterogeneity and cell composition
D1 uses cord blood, D2/D3 peripheral blood, and there was no cell-composition adjustment.
Please: if basic differentials are available, adjust for leukocyte fractions (e.g., neutrophil/lymphocyte proportions) as covariates; otherwise, elevate this limitation in the Discussion regarding reduced sensitivity to detect small effects.

Reviewer 2 Report

Comments and Suggestions for Authors

This is a well-executed, carefully written observational epigenetic study examining SLC6A4 methylation in preterm infants receiving Kangaroo Mother Care (KMC) versus standard care. The research is valuable. However, the methodological constraints fundamentally limit the interpretability of the results.

The cohort includes 51 infants in the KMC group and 24 in the non-KMC group, with substantial clinical variability. Since DNA methylation differences in this setting are very small, the sample size is almost certainly insufficient to reliably detect CpG-specific differences. Although the authors acknowledge this limitation, its implications for the validity and interpretability of the findings are likely under-addressed.

Across the 13 CpG sites analyzed, most methylation levels fall within very low ranges, with the exception of CpG12–13. Such small absolute values make it difficult to detect meaningful or reproducible methylation differences.