Cord Blood Metabolome and BMI Trajectory from Birth to Adolescence: A Prospective Birth Cohort Study on Early Life Biomarkers of Persistent Obesity

Round 1

Reviewer 1 Report

Dear Authors,

The manuscript presents a study of cord blood metabolome and BMI trajectory from birth to adolescence to evaluate putative early life biomarkers of obesity. The work provides interesting, new findings based on substantial experimental work performed in a large cohort. The authors demonstrated distinctive longitudinal BMI trajectories and identified cord plasma metabolites in relevant biological pathways associated with early-OWO.

I have some minor considerations regarding missing elements in the manuscript.

The manuscript is mainly composed of a large results section, but obtained results are discussed pretty shortly. A manuscript may benefit from extending the discussion section to explain the putative clinical relevance of obtained results (TAGs, DAGs, CE relations with OWO) and propose biochemical mechanisms resulting from obtained data. A critical authors view of described topic may be of value.

Did all mothers give birth by cesarean section? Thus, it is worth mentioning in the discussion section the role of the microbiome in obesity development.

There are some shortcuts without the explanation in the manuscript (for example, the full name of R WGCNA). Please, explain the shortcuts.

Please provide the company for apparatus used in the study and the name of the software that generated presented images (for example, heatmaps)

Please improve the quality of the Figures of heatmaps. The titles are hard to read.

In my opinion, the abbreviations for metabolite types should also be provided in the description of Figures to make them more transparent for readers.

The shortcuts should not be explained again in the text of the manuscript. Please check the text carefully.

There is some typo in the manuscript. Please check the text carefully. Also, the shortcut "et al." is not always present.

Author Response

Please find our response (in-line in red) to Reviewer 1's comments in the attached file below.

Author Response File: Author Response.docx

Reviewer 2 Report

The topic presented in this article is really interesting. The paper submitted by Tingyi Cao and colleagues reported that a range of triacylglycerols and diacylglycerols were negatively associated with early-OWO and cholesterol esters were positively associated with early-OWO. The study shows that metabolites in cord blood metabolomic profile could assist in the identification of early-OWO. Also, I agree that a strength of their study is to use of cord blood metabolomics for examining children’s BMI trajectories.

Major comments

Lines 166 and 203. It is well known that BMI values may not well identify individuals with high-fat content, i.e., the truly obese individuals. The authors should try to discuss this potential limitation of their study and confirm the reliability of the results. The authors should use the z-score for the analyses instead of crude BMI in the BMI, which in paediatrics is never used.

Lines 166 and 203. Using longitudinal data models there is a great opportunity to estimate cluster-specific effects for understanding not only the cluster but also the interindividual variability in longitudinal responses. The authors should try to use, as an alternative, well-established approaches in analysing longitudinal data(eg Mixed Effect Regression Models or Generalized Estimating Equations).  

Line 246. Small-for-gestational-age (SGA) birth has been associated with increased adiposity in childhood. Additionally, birth weight has been associated with cord blood metabolomics. Although authors have these variables, they have not adjusted the models for them. Could the authors provide further information regarding the impact of these variables on the models? Could this adjustment affect BMI association with TAG?

Lines 135-137. Please, provide a more detailed description about what analysis has been followed for identifying that 2 is the best number of clusters? How do the authors find the optimum hyperparameters of the model?

Lines 167-170. Please, provide a more detailed description of what analysis has been followed for identifying the optimum parameters (e.g. minimum module size)?

Lines 398-401. Could the authors check for more recent publications on cord blood metabolomic profile as a biomarker of predicting overweight/obesity in childhood growth curves. Could the authors examine the potential predictive ability of the cord blood metabolic profiles in other cohorts or maybe in the examined population?  

There are also a lot of studies that have shown that rapid growth is a risk factor of later overweight/obesity?  Is cord blood metabolomics associated with rapid growth? Does rapid growth affect BMI trajectories?

Taking into account that the authors had a multi-ethnic US cohort, I suggest adding nutritional info of the mother and the infants. Results like the TAGs and DAGs could be related to the autochthon diet.

Family nutrition and physical activity are very important risk factors for childhood obesity. Can these factors affect the BMI trajectories during childhood?

Minor comments:

Line 24: Could the authors provide the “background” of the problem in the abstract?

Line 177. It is very difficult to read the y-axis (the names of metabolic traits) of figure 2. It would be easier for the reader to see only the metabolites that have p-value or/and FDR lower a threshold (eg 0.05). 

Lines 177 and 288. The authors should need to check the coloured clusters/categories in figures 2 and 4. Valine is a branched-chain amino acid (BCAA). BCAA could be in an additional cluster. In the heatmaps, valine is in the cluster named “others”.

Line 190. The authors could add in the table caption information related to model adjustment.  I suggest this for all the captions of the manuscript and supplementary material.

Lines 228-231. It is very difficult to read the y-axis (metabolic traits) of figure 4. It would be easier for readers to see only the metabolites that have p-value or/and FDR lower a threshold (eg 0.05). The overall results can go to supporting information as a table.  

Lines 130-134.  Please provide a detailed explanation in the methods sections about the cut off (eg BMI z-score >2) considered to characterize early-onset overweight or obesity (early-OWO), late-onset overweight or obesity (late-OWO). 

Line 319. Could the authors please check table 3? There are odds ratios and CIs that are the same? Why does this happen? Are these highly correlated metabolic traits?  Also, I strongly suggest to the authors present this table using a forest plot.

Line 140: Please provide explain what is LOWESS.

Lines 426-427  Add a reference for this affirmation. Which birthweight (“with greater birthweight”) did you consider as larger infants  Could the author evaluate this using the existing dataset?

Line 404. Add a reference for this affirmation

The author could check for updating the introduction with maybe more recent publications on cord blood metabolomics and childhood obesity.

Author Response

Please find our response (in-line in red) to Reviewer 2's comments in the attached file below.

Author Response File: Author Response.docx