Associations Between Human Milk Oligosaccharides and Maternal Nutrition: Latvian Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study by Aumeistere et al. for the first time explored the associations between maternal dietary intake, supplement use, and the concentration of human milk oligosaccharides (HMOs) in a cohort of 68 exclusively breastfeeding women in Latvia—a population previously underrepresented in HMO research. The authors collected rich, paired data including 24-hour pooled human milk samples, 72-hour food diaries, and detailed sociodemographic and anthropometric questionnaires, reflecting an impressive logistical effort for a national pilot study. Their dietary assessment went beyond mere quantification: nutrient intakes were benchmarked against Nordic recommendations, revealing widespread inadequacies in vitamin A, C, folate, and iodine—thereby grounding their findings in public health relevance. Notably, the study examined not only macronutrients but also specific dietary behaviors (e.g., dairy avoidance, preference for “zero sugar” products) and common supplement use (vitamin D, calcium, omega-3, zinc), linking these nuanced exposures to individual HMO profiles—a level of granularity rarely seen in the literature. The authors also conducted inter-HMO correlation analyses, offering insights into potential biosynthetic relationships among the eight major oligosaccharides. Methodologically, their approach was sound: all continuous variables were appropriately reported as medians with ranges, Spearman’s rank correlation was used to assess associations, and group comparisons (e.g., supplement users vs. non-users) relied on non-parametric tests consistent with the skewed distribution of HMO data. Crucially, the authors exercised restraint in interpretation, highlighting only a few robust associations (such as calcium use with higher 2′-FL or vitamin D with elevated 6′-SL), which suggests cautious handling of multiple testing despite the absence of formal correction. This context further underscores the value of the study, as it provides crucial baseline data for future regional nutritional interventions and human milk science, despite its limited sample size. Altogether, this work provides a valuable, methodologically appropriate foundation for future research on nutritional modulation of HMOs in diverse populations.

comments: 

Including all 14 covariates from Tables 1–2 would risk overfitting (for 68 sample size), but a focused adjustment for core biological and temporal factors would enhance confidence in the observed links without compromising model stability. For the 2–3 most biologically compelling associations (e.g., calcium supplementation and 2′-FL; dairy restriction and 2′-FL), consider adding a sensitivity analysis using a parsimonious multivariable model (e.g., quantile regression or robust linear regression on log-transformed HMOs) that adjusts for 3–5 key confounders—such as secretor status (defined by 2′-FL < 200 mg/L), maternal age, infant age (lactation duration), total energy intake, and BMI.

The authors should carefully proofread the manuscript for grammatical accuracy. to avoid typos and grammar mistakes, such as line 21  Participants had insufficient intake of vegetables, fruits, berries, milk & dairy products, and fish, leading to vitamin A, vitamin C, folate, and iodine intakes lower than are recommended for lactating women.

The formatting of p-values is inconsistent across the manuscript, such as A p-value of ≤0.050 in line 186, p > 0.05 in line 237, p < 0.05 in line 378

Author Response

We would like to thank the Reviewer for the comments and suggestions provided to improve the manuscript (ID: nutrients-4070124). Below, we have included a point-by-point description of how each comment was addressed in the manuscript.  Original comments in bold, responses in italic. 

Comment 1. Including all 14 covariates from Tables 1–2 would risk overfitting (for 68 sample size), but a focused adjustment for core biological and temporal factors would enhance confidence in the observed links without compromising model stability. For the 2–3 most biologically compelling associations (e.g., calcium supplementation and 2′-FL; dairy restriction and 2′-FL), consider adding a sensitivity analysis using a parsimonious multivariable model (e.g., quantile regression or robust linear regression on log-transformed HMOs) that adjusts for 3–5 key confounders—such as secretor status (defined by 2′-FL < 200 mg/L), maternal age, infant age (lactation duration), total energy intake, and BMI. 

Response 1 We fully agree that adjusting for key biological and temporal factors is important, particularly given the modest sample size. When selecting the type of data processing, it was important for us to understand the strength of the relationship; therefore, we now used partial Spearman correlations to assess associations while controlling for the covariate – time postpartum. We have added additional information to the manuscript to clarify where observed associations regarding HMO and dietary factors were influenced by covariate – time postpartum. 

Comment 2 The authors should carefully proofread the manuscript for grammatical accuracy. to avoid typos and grammar mistakes, such as line 21  Participants had insufficient intake of vegetables, fruits, berries, milk & dairy products, and fish, leading to vitamin A, vitamin C, folate, and iodine intakes lower than are recommended for lactating women. 

Response 2 We used Grammarly to correct grammatical mistakes and improve the readability and language of the manuscript. 

Comment 3 The formatting of p-values is inconsistent across the manuscript, such as A p-value of ≤0.050 in line 186, p > 0.05 in line 237, p < 0.05 in line 378 

Response 3 We have corrected p-values in the manuscript, indicating them with three digits after the decimal point. In a few places in the text, individual p‑values were not listed; instead, these results were collectively reported as p > 0.050 or p < 0.050. 

Reviewer 2 Report

Comments and Suggestions for Authors

I have read this paper with great interest, with a background on perinatal clinical research (including aspects of human milk and its composition), and value the effort as reported. I obviously agree that HMO are a relevant component of human milk, with increasing interests in this specific topic, including the potential associations between maternal diet and HMO composition in the milk.

I understand the ‘weighted’ approach for the 24 h milk collection, but it seems that eg for- versus hindmilk or other aspects have not been standardized, not collection technique. While pragmatic, this is a limitation. I would at least urge the authors to better describe in details the collection technique, and perhaps add (translated version) the instructions as an addendum. At present, I rather understand that a ‘pooled’ sample, reflected 24 h human milk composition was collected ?

The bioanalysis reads adequate. How have HMO been ‘selected’ to be quantified?

How representative was the cohort, when weighted to the Latvian pregnancy population (primigravida, age, socio-economics, smoking, type of delivery etc). Could there be biases ?

My main concern relates to the rather ‘causal’ language used to link dietary intake to human milk composition. In my opinion, the current design is not able to generate causal data, but rather association type of findings. I therefore would suggest the reconsider the wording used.

What ‘zero sugars’ have been considered, stevia, aspartame, others ?

Line 34: HM has more context than ‘energy’, perhaps rephrase ?

Author Response

We would like to thank the Reviewer for the comments and suggestions provided to improve the manuscript (ID: nutrients-4070124). Below, you will find a point-by-point description of how each comment was addressed in the manuscript.  Original comments in bold, responses in italic. 

Comment 1 I understand the ‘weighted’ approach for the 24 h milk collection, but it seems that eg for- versus hindmilk or other aspects have not been standardized, not collection technique. While pragmatic, this is a limitation. I would at least urge the authors to better describe in details the collection technique, and perhaps add (translated version) the instructions as an addendum. At present, I rather understand that a ‘pooled’ sample, reflected 24 h human milk composition was collected ? 

Response 1 More detailed information on the human milk sampling is already available in the informed consent template deposited in the data repository (https://doi.org/10.71782/DATA/RGML1E; see Data Availability Statement). In response to the reviewer’s suggestion, we will additionally upload the full Human Milk Sampling Instruction that was provided to each participant to the same repository. 

Comment 2 The bioanalysis reads adequate. How have HMO been ‘selected’ to be quantified? 

Response 2 A detailed and comprehensive literature review was conducted, focusing specifically on HMOs that have been identified using UHPLC coupled with fluorescence detection. In selecting target compounds, it was essential to prioritize HMOs that could be analyzed under uniform chromatographic conditions – namely, the same column type, consistent column temperature, and without the need for excessive solvent use. Practical considerations also played a key role: the chosen reference standards needed to be readily accessible, of high purity, and available at a reasonable cost. The HMOs selected for this study – 2′‑Fucosyllactose (2′‑FL), 6′‑Sialyllactose (6′‑SL), 3‑Fucosyllactose (3′‑FL), 3′‑Galactosyllactose (3′‑GL), 6′‑Galactosyllactose (6′‑GL), Lacto‑N‑difucohexaose I (LNDFH I), Lacto‑N‑difucohexaose II (LNDFH II), and Lacto‑N‑neotetraose (LNnT) – are among the most frequently reported HMOs in the scientific literature. This supports a more robust interpretation of the results obtained from this study. 

Comment 3 How representative was the cohort, when weighted to the Latvian pregnancy population (primigravida, age, socio-economics, smoking, type of delivery etc). Could there be biases ? 

Response 3 Participation in the study was voluntary, and the cohort was not stratified or weighted to match the sociodemographic structure of the Latvian pregnancy population. As a result, the sample may differ from the national population in characteristics such as parity, age, socio‑economic status, smoking behavior, and mode of delivery. These differences could introduce selection bias, and we acknowledge in the Conclusion section that the cohort should not be considered fully representative. More human milk samples and dietary data need to be evaluated. 

Comment 4 My main concern relates to the rather ‘causal’ language used to link dietary intake to human milk composition. In my opinion, the current design is not able to generate causal data, but rather association type of findings. I therefore would suggest the reconsider the wording used. 

Response 4 Terms that could be interpreted as causal have been replaced with neutral phrasing such as “was associated with”, “was linked to,” or “showed a relationship with.”  

Comment 5 What ‘zero sugars’ have been considered, stevia, aspartame, others ? 

Response 5 In our questionnaire, the term “zero sugar” was used in the way it commonly appears on food labels and in consumer marketing, and participants were asked only whether they preferred products in which sugar is replaced with sweeteners. It was not specified in the questionnaire which types of sweeteners this might include. As a result, participants’ responses likely encompassed a broad spectrum of sweeteners that can be used for “zero sugar” products. These may include non‑sugar sweeteners (e.g., acesulfame K, aspartame, saccharin, sucralose, steviolglycosides), low‑calorie sugars, sugar alcohols (polyols). Because the questionnaire did not differentiate between these categories, the variable “preference for zero‑sugar products” reflects participants’ general attitudes toward zero-sugar products rather than intake of any specific type of sweetener. We have pointed this out as a limitation in the Discussion section (L387-394). 

Comment 6 Line 34: HM has more context than ‘energy’, perhaps rephrase ? 

Response 6 Line 34 has been rephrased.