Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript investigates the effects of standard, high-carbohydrate, and low-carbohydrate/high-fat diets using lard or coconut oil in an AOM/DSS mouse model of colitis-associated colorectal carcinogenesis. The study reports that the low-carbohydrate high-fat diet groups had higher body weight, relatively preserved colon length, reduced tumor number and tumor size, increased lymphocyte infiltration per tumor, and altered gut microbiome composition/diversity. The topic is relevant because diet-microbiome-inflammation interactions are important in colorectal cancer biology. However, several major experimental and analytical limitations prevent strong conclusions in its current form.

The study is potentially interesting, but the current conclusions are stronger than the data support. The manuscript requires substantial revision, especially regarding statistical analysis, dietary control, interpretation of lymphocyte infiltration, and histopathological validation.

Major Comments

1. The high-fat diets contain substantially higher caloric density than the standard and high-carbohydrate diets. The authors should provide food intake data and, ideally, discuss whether pair-feeding or isocaloric dietary design would be needed to support their conclusion.
2. The study includes only n=4 controls and n=6 per AOM/DSS diet group. The authors should reanalyze the data using ANOVA, followed by corrected post hoc testing. The statistical methods currently state only two-tailed Student’s t-test.
3. The manuscript reports tumor number and total tumor size, but does not provide histopathological scoring of lesions, such as inflammation, dysplasia, adenoma, adenocarcinoma, or tumor grade. In an AOM/DSS model, macroscopic lesions alone are insufficient to establish effects on colorectal carcinogenesis. The authors should include histological classification and representative H&E images of tumors from each diet group, preferably scored by a blinded pathologist.
4. The manuscript interprets H&E-based lymphocyte infiltration as tumor-associated immune response. However, H&E cannot reliably distinguish CD8 T cells, CD4 T cells, B cells, regulatory T cells, macrophages, or other immune populations. This is especially important because the biological interpretation depends on whether the infiltrate is antitumor or inflammatory. The authors should either perform immunohistochemistry or clearly state that the current analysis only reflects morphologic inflammatory cell infiltration, not defined antitumor lymphocyte immunity.
5. Because the HFL and HFC groups had fewer tumors, dividing lymphocyte area by tumor number may artificially inflate “per-tumor” infiltration. Total infiltration area was actually highest in the HCD group, while per-tumor infiltration was higher in HFL/HFC. I was quite concerned that authors chose to analyze data this way. The authors should present this cautiously and consider additional normalization, such as infiltration per tumor area, per mucosal area, or per lesion grade.
6. To support the proposed interpretation, the authors should consider measuring serum glucose, insulin, triglycerides, cholesterol, ketone bodies, inflammatory cytokines, colon cytokine markers, and/or fecal bile acids or short-chain fatty acids. These missing data, if absent or if they are unable to perform, should be acknowledged as limitations. Also, coconut oil-derived fatty acids, especially lauric acid, have reported antitumor, antimicrobial, and immunomodulatory effects. This should be incorporated into the Discussion and considered as a limitation.

Minor Comments

1. All figure legends should specify the exact n per group and whether each dot represents an individual mouse or technical replicate.
2. The manuscript should clarify whether tumor counting, tumor measurement, and histological scoring were performed blinded.
3. The control group did not receive AOM/DSS and was maintained on standard diet only. There are no non-AOM/DSS controls for HCD, HFL, or HFC diets. This limits interpretation of diet-specific microbiome changes independent of carcinogen/colitis treatment. 

Author Response

We sincerely thank the reviewer for the careful evaluation of our manuscript and for the constructive comments. We agree that several interpretations in the original manuscript were stronger than supported by the available data. We have revised the manuscript substantially by reanalyzing the quantitative data using one-way ANOVA with Tukey's multiple-comparison test, toning down causal and immunological interpretations, clarifying the non-isocaloric diet design, specifying analyzed sample sizes in figure legends, and expanding the limitations.

Major Comments

1. The high-fat diets contain substantially higher caloric density than the standard and high-carbohydrate diets. The authors should provide food intake data and, ideally, discuss whether pair-feeding or isocaloric dietary design would be needed to support their conclusion.

-> Response: We agree. Food intake was not recorded in the present experiment, and the high-fat diets had higher caloric density than the SD and HCD diets. We have added this information to the Materials and Methods and Discussion, and we now state that pair-feeding or an isocaloric design would be required to determine whether carbohydrate restriction or fat source independently affects colorectal tumorigenesis. We also toned down the conclusion accordingly.

2. The study includes only n = 4 controls and n = 6 per AOM/DSS diet group. The authors should reanalyze the data using ANOVA, followed by corrected post hoc testing. The statistical methods currently state only two-tailed Student's t-test.

-> Response: We reanalyzed the quantitative data using one-way ANOVA followed by Tukey's multiple-comparison test. The Statistical Analysis section and figure legends have been revised. Importantly, the reanalysis showed that some originally stated pairwise differences did not remain significant after correction. We revised the Results accordingly, including the lymphocyte infiltration analysis.

3. The manuscript reports tumor number and total tumor size, but does not provide histopathological scoring of lesions, such as inflammation, dysplasia, adenoma, adenocarcinoma, or tumor grade.

-> Response: We agree that macroscopic lesions alone are insufficient to establish histological progression of colorectal carcinogenesis. Formal lesion grading was not performed in the present analysis. We therefore revised the Results and Discussion to describe these findings as macroscopic tumor burden, not definitive histological progression, and added the absence of inflammation/dysplasia/adenoma/adenocarcinoma/tumor-grade scoring as a limitation.

4. The manuscript interprets H&E-based lymphocyte infiltration as tumor-associated immune response. However, H&E cannot reliably distinguish CD8 T cells, CD4 T cells, B cells, regulatory T cells, macrophages, or other immune populations.

-> Response: We agree. We revised this section throughout the manuscript. The analysis is now described as H&E-based morphologic lymphocyte-like inflammatory cell infiltration, not defined antitumor lymphocyte immunity. We also state that immunohistochemistry or flow cytometry would be required to identify immune-cell subsets and determine the biological meaning of the infiltrate.

5. Because the HFL and HFC groups had fewer tumors, dividing lymphocyte area by tumor number may artificially inflate "per-tumor" infiltration.

-> Response: We agree with this concern. We now present the per-tumor analysis as exploratory and explicitly state that normalization by tumor number can inflate values in groups with fewer tumors. After ANOVA with Tukey correction, per-tumor infiltration did not show statistically significant differences among diet groups; the Results have been revised accordingly. We also added that normalization to tumor area, mucosal area, or lesion grade would be needed in future studies.

6. To support the proposed interpretation, the authors should consider measuring serum glucose, insulin, triglycerides, cholesterol, ketone bodies, inflammatory cytokines, colon cytokine markers, and/or fecal bile acids or short-chain fatty acids. These missing data should be acknowledged as limitations. Also, coconut oil-derived fatty acids, especially lauric acid, have reported antitumor, antimicrobial, and immunomodulatory effects.

-> Response: We agree. These metabolic, inflammatory, and microbial metabolite measurements were not performed in the present study. We added these missing measurements as limitations. We also incorporated a discussion of coconut oil-derived medium-chain fatty acids, particularly lauric acid, and noted that the findings in the HFC group may reflect effects specific to coconut oil-derived fatty acids as well as macronutrient composition.

 

Minor Comments

1. All figure legends should specify the exact n per group and whether each dot represents an individual mouse or technical replicate.

-> Response: We revised the figure legends to state that dots or points represent individual mice or individual fecal samples, not technical replicates. We also added allocated and analyzed sample-size information where applicable.

2. The manuscript should clarify whether tumor counting, tumor measurement, and histological scoring were performed blinded.

-> Response: Blinding was not documented in the original experimental record. We added this clarification to the Materials and Methods and acknowledged it as a limitation.

3. The control group did not receive AOM/DSS and was maintained on standard diet only. There are no non-AOM/DSS controls for HCD, HFL, or HFC diets. This limits interpretation of diet-specific microbiome changes independent of carcinogen/colitis treatment.

-> Response: We agree. We added this issue to the limitations and state that the lack of non-AOM/DSS controls for HCD, HFL, and HFC limits interpretation of diet-specific microbiome effects independent of AOM/DSS-induced colitis and carcinogenesis.

 

Reviewer 2 Report

Comments and Suggestions for Authors

This study investigates the effects of standard diet, high-carbohydrate diet, lard-based low-carbohydrate high-fat diet, and coconut oil-based low-carbohydrate high-fat diet on tumorigenesis, immune infiltration, and gut microbiome in an AOM/DSS-induced mouse model of colitis-associated colorectal cancer. The topic addresses the clinical question of dietary nutrition and colorectal cancer prevention, with certain scientific value. The study has a complete design covering physiological, pathological, and microbiome dimensions. However, there are critical flaws in experimental design, statistical methods, mechanistic depth, and result interpretation. Major Revision is recommended, and the manuscript can be considered for re-review after revision.

 

1. Section 4.2 Animal grouping and diet formulation, Table 1 (lines 410–411): Among the four experimental groups, the two high-fat diets (HFL, HFC) feature two simultaneous variables: low carbohydrate and high fat. Compared with the SD and HCD groups, both carbohydrate proportion and fat proportion change drastically. Therefore, the observed differences in tumorigenesis, immunity, and microbiome cannot be unambiguously attributed to carbohydrate restriction, increased total fat, or different fat sources. There is an essential flaw in the causal attribution of the core research question.If additional experiments are not feasible, explicitly acknowledge this design limitation in the Discussion section, emphasizing that this study only verifies the overall effect of the "low-carbohydrate high-fat dietary pattern" and cannot disentangle the independent effects of carbohydrate and fat. Ideally, supplement additional experimental groups: e.g., normal-carbohydrate high-fat (lard/coconut oil) groups, or adjust the proportion of a single macronutrient under isocaloric conditions to separate the independent effects of carbohydrate and fat.

 

2. Section 4.5 Statistical analysis (lines 462–466): The study includes 5 experimental groups, and all pairwise comparisons use Student’s t-test. Repeated t-tests in multi-group designs significantly increase the probability of type I error (false positives), which does not comply with statistical norms in biomedical research. In addition, longitudinal repeated-measures data such as body weight and microbiome are not analyzed with appropriate methods, so the interaction effect between time and diet cannot be evaluated.Use one-way ANOVA for overall comparison among multiple groups, and perform Tukey’s post-hoc test for pairwise comparisons only when the global p-value < 0.05. Apply repeated-measures ANOVA for longitudinal data across time points (body weight, microbiome diversity, genus abundance) to evaluate the effects of diet, time, and their interaction. Recalculate all statistics throughout the manuscript and update significance labels in all figures.

 

3. Section 4.4 Gut microbiome analysis (lines 450–461); microbiome-related paragraphs in Discussion (lines 334–370): Only V3-V4 region 16S rRNA sequencing is used, which only achieves genus-level resolution and cannot identify key pathogenic bacteria closely related to colorectal cancer. Meanwhile, the Discussion extensively speculates that the microbiota regulates tumorigenesis via short-chain fatty acids, bile acid metabolism, and intestinal barrier function, but no microbial metabolites or barrier function indicators are measured. All mechanistic inferences are based on previous literature without direct experimental evidence from this study. Supplement functional prediction analysis (e.g., PICRUSt2) to preliminarily indicate inter-group differences in metabolic pathways.If conditions permit, add targeted detection of fecal short-chain fatty acids and bile acid profiles, or intestinal barrier indicators such as colonic tight junction proteins and serum LPS, to validate mechanistic hypotheses. Weaken definitive mechanistic conclusions in the Discussion, explicitly mark statements as speculative, and list the lack of metabolomic data as a study limitation.

 

4. Results section 2.4 Microbiome composition analysis (lines 196–226); Discussion lines 351–356: The study finds that high-fat diet groups have significantly higher abundance of recognized pro-inflammatory/pro-carcinogenic taxa such as Proteobacteria, Escherichia/Shigella, and Helicobacter, with more pronounced dysbiosis, yet their tumor number and total size are significantly lower than those in the SD and HCD groups. This core finding contradictory to current understanding is only briefly mentioned in the Discussion as "microbial alterations may not directly parallel tumor burden", without in-depth mechanistic discussion. This is the most critical logical flaw of the study.Add a dedicated paragraph in the Discussion to systematically analyze potential mechanisms underlying this paradox: ① The tumor-suppressive effect of impaired glycolysis in tumor cells under low-carbohydrate conditions outweighs the pro-carcinogenic effect of microbial dysbiosis; ② Medium-chain fatty acids in coconut oil exert direct anti-tumor and immunomodulatory effects; ③ Ketogenesis induced by high-fat diet exerts anti-tumor effects.Integrate the finding of increased lymphocyte infiltration to propose an integrated hypothesis that "microbial dysbiosis activates anti-tumor immune responses" to explain the experimental phenomena. Explicitly state that dietary regulation of colorectal cancer is a combined effect of metabolism, immunity, and microbiota, and tumor risk cannot be judged solely by microbial composition.

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

We sincerely thank the reviewer for the careful and constructive evaluation of our manuscript. We agree that the original version required clearer acknowledgment of the dietary design, more appropriate statistical framing, and a more cautious interpretation of microbiome-related findings. We have revised the manuscript substantially by emphasizing that the study compares dietary patterns rather than isolated macronutrient effects, by clarifying the statistical approach, by weakening mechanistic interpretations based on 16S rRNA sequencing, and by adding a dedicated discussion of the apparent paradox between dysbiosis and lower macroscopic tumor burden in the low-carbohydrate high-fat diet groups.

 

1. The HFL and HFC diets include two simultaneous variables: low carbohydrate and high fat. Therefore, the observed effects cannot be unambiguously attributed to carbohydrate restriction, increased fat, or fat source.

-> Response: We agree. We revised the Abstract, Introduction, Materials and Methods, Discussion, and Limitations to state explicitly that the HFL and HFC groups represent low-carbohydrate high-fat dietary patterns and that the present design cannot disentangle the independent effects of carbohydrate restriction, increased total fat, fat source, or caloric density. We also added that isocaloric, pair-fed, normal-carbohydrate high-fat, or single-macronutrient-controlled experimental groups would be required to isolate these effects.

2. The statistical analysis used repeated Student's t-tests in a multi-group design, and longitudinal repeated-measures data were not analyzed with appropriate methods.

-> Response: We agree with the concern regarding type I error. The endpoint data were reanalyzed using one-way ANOVA followed by Tukey's multiple-comparison test when the global ANOVA p-value was < 0.05, and the Results, Statistical Analysis section, and figure legends were revised accordingly. For body weight, only final body weight was used for formal endpoint comparison; longitudinal body-weight patterns were interpreted descriptively. For the longitudinal microbiome data, the small cohort size, missing samples, and exploratory 16S design limited robust repeated-measures ANOVA or mixed-effects modeling. We therefore revised the Results and Statistical Analysis sections to present longitudinal body-weight and microbiome findings descriptively and to avoid claiming definitive diet-by-time interaction effects. We also added this as a limitation.

3. The microbiome analysis used only V3-V4 16S rRNA sequencing, and the Discussion speculates about short-chain fatty acids, bile acid metabolism, and barrier function without direct measurements. Functional prediction such as PICRUSt2 or metabolite/barrier analyses should be considered.

-> Response: We agree. PICRUSt2, fecal short-chain fatty acid analysis, bile acid profiling, serum LPS, tight-junction protein analysis, and other barrier-function assays were not performed in the present study. We revised the microbiome Results, Methods, Discussion, and Limitations to clarify that V3-V4 16S rRNA sequencing provides compositional profiling with limited species-level and functional inference. We also explicitly state that proposed links to microbial metabolites, bile acid metabolism, epithelial barrier function, and tumor-promoting pathways remain speculative and require future metagenomic, metabolomic, functional-prediction, and barrier-function studies.

4. The high-fat diet groups showed increased Proteobacteria, Escherichia/Shigella, and Helicobacter, but lower tumor number and size. This apparent contradiction requires deeper discussion.

-> Response: We agree that this is a central interpretive issue. We added a dedicated paragraph in the Discussion addressing this paradox. We now discuss several possible explanations: reduced carbohydrate availability may limit glucose-dependent tumor growth; coconut oil-derived medium-chain fatty acids, including lauric acid, may have antimicrobial, immunomodulatory, or antitumor effects; low-carbohydrate high-fat feeding may promote ketogenesis; and inflammatory cell infiltration may reflect immune activation, although H&E staining cannot determine whether this is antitumor immunity or nonspecific inflammation. We emphasize that colorectal tumor risk cannot be inferred solely from microbial composition and that dietary modulation of tumorigenesis likely reflects combined effects of host metabolism, immune responses, and microbial ecology.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed the major concerns. The revised manuscript is now an exploratory study of dietary-pattern associations with macroscopic tumor burden and gut microbiome changes in the AOM/DSS model.

Minor suggestions:

1) Suggested title: Associations of low-carbohydrate high-fat dietary patterns with colorectal tumor burden and gut microbial dynamics in an AOM/DSS mouse model.

2)  This sentence is still a little strong: “the increased carbohydrate availability may have created metabolic conditions favorable for tumor growth”. Consider saying "The higher tumor burden in the HCD group may be consistent with a metabolic environment permissive for tumor growth, but this interpretation remains speculative because metabolic parameters and food intake were not measured."

3) Clarify or remove significance stars in microbiome diversity Figure 5. The text says the microbiome diversity results are descriptive because repeated-measures analysis was limited by small sample size, but Figure 5 still says “*p < 0.05 where indicated.”

 

Author Response

The authors have addressed the major concerns. The revised manuscript is now an exploratory study of dietary-pattern associations with macroscopic tumor burden and gut microbiome changes in the AOM/DSS model.

- Response: We sincerely thank the reviewer for recognizing the revisions and for the constructive minor suggestions. We agree that the exploratory framing should be reflected consistently throughout the manuscript.

Minor suggestion 1: Suggested title: Associations of low-carbohydrate high-fat dietary patterns with colorectal tumor burden and gut microbial dynamics in an AOM/DSS mouse model.

- Response: We agree and have changed the title as suggested.

Minor suggestion 2: This sentence is still a little strong: "the increased carbohydrate availability may have created metabolic conditions favorable for tumor growth". Consider saying "The higher tumor burden in the HCD group may be consistent with a metabolic environment permissive for tumor growth, but this interpretation remains speculative because metabolic parameters and food intake were not measured."

- Response: We agree. We replaced the original sentence with the suggested more cautious wording to avoid overinterpretation.

Minor suggestion 3: Clarify or remove significance stars in microbiome diversity Figure 5. The text says the microbiome diversity results are descriptive because repeated-measures analysis was limited by small sample size, but Figure 5 still says "*p < 0.05 where indicated."

- Response: We agree. We revised the Figure 5 legend to clarify that the asterisks in panel A indicate nominal, unadjusted pairwise comparisons at individual time points and are provided only as descriptive markers. We also state explicitly that, because of the small sample size and incomplete longitudinal matching, no formal repeated-measures or diet-by-time interaction analysis was performed.

 

Sincerely,

Seun Ja Park, MD, PhD 

Jeonghoon Heo, PhD

Reviewer 2 Report

Comments and Suggestions for Authors

The author replied to my question very carefully, and I have no further questions.

Author Response

The author replied to my question very carefully, and I have no further questions.

-  We sincerely thank the reviewer for the careful evaluation of our manuscript and for the positive comments. We greatly appreciate your thoughtful review and are pleased that our revisions have satisfactorily addressed your concerns.