Effects of Whole-Grain and Sugar Content in Infant Cereals on Gut Microbiota at Weaning: A Randomized Trial

Round 1

Reviewer 1 Report

The authors present the results of study to assess whether infants receiving cereals with or without whole grains (and consequently different amounts of simple sugars) exhibit differences in the establishment of their gut microbiota. Their key finding is that whole grain cereals may lead to a reduction in the percent abundance of Proteobacteria (notably Escherichia) in the developing infant gut microbiota and increases in a handful of other taxa.

The design of the study is appropriate to a study of this type, and the data analysis is mostly appropriate, though some changes to the data processing procedures, analytical methods, and presentation  of results may improve the report. My comments below address these points.

Overall comments:

This paper appears to use the OTU approach (though this is not adequately described in the methods). This method of clustering sequences and then assigning taxonomy to representative sequences within clusters, while not wrong, has largely been supplanted by amplicon sequence variants, such as those yielded by the DADA2 pipeline. While neither is perfect, ASVs should be better at avoiding some of the pitfalls associated with grouping different taxa based on an arbitrary % sequence identity. This could prove interesting in the developing infant gut microbiota, where Bifidobacteria make up a large percentage of the community, but contain many different related taxa. Likewise, aggregating relative abundances at the genus and phylum levels will mask interesting patterns that may emerge within important genera like Bifidobacterium and Bacteroides, which have several members with very different preferences for different nutrient sources. Analyzing the data at the ASV level is almost always more informative than at the phylum and genus levels in a study of this type.

Comments on methods:

1. ANCOVA is not typically appropriate for relative abundance data, which often exhibit zero-inflation and greater variances with greater mean values. Methods for count-based data (controlling for sequencing depth), or models of variance-stabilized counts, or Rivera-Pinto's 'balances' method might all be more appropriate for statistical tests of differences in relative abundance.
2. Likewise, though PCA is used by some authors to investigate the microbiota, it is usually considered to be inappropriate for relative abundances. A more accepted approach is to perform principal coordinates analysis on a dissimilarity or distance metric. The authors use PERMANOVA on Bray-Curtis dissimilarities (assuming this is the default they mention), so a natural solution would be to present PCoA of the same dissimilarity, which would provide a visualization to directly accompany their tests.
3. The correlations need to be adequately described. Are they Spearman's or Pearson's?

Comments on results:

1. The PCA and PERMANOVA results need more information presented to interpret them fully. First, the R-squared (estimated variance explained) should be presented for the factors in the PERMANOVA. Second the PCA results currently present the groups as though all the points are independent. Some effort should be made to show how individuals are moving over time. Are they all moving a similar distance in the space? Is their change highly variable? Is their change dependent on treatment. Tests applied to the dissimilarities between the two samples of an individual may be informative.
2. Figure 2 on page 8 - in the legend, the authors should be explicit that normal faced taxa represent differences over time. Also, Figure 2 could benefit from a different presentation. The current design summarizes key results, but is entirely redundant with Tables 4 and 5, and does not present any perception of effect size, only significance.
3. Figures 3 and 4 on page 11 could be better presented. Figure 3 is currently arranged to better depict variation between individuals than change over time or treatment differences, and it is unclear why Figure 4 has a different arrangement. The authors would be better served by graphical representations of the key differences they observe in specific taxa. Also, does figure 4 present means? The legend needs to indicate this clearly.
4. Figure 6 needs a legend - it is currently a repeat of figure 5's legend.
5. The correlation section isn't currently contributing much to the results, and the authors need to be clearer in both the methods and results about which groups and timepoints are in each set presented. It contains 3 large figures that are discussed in very minor ways, and the results aren't surprising (e.g., the dominant Bifidobacterium being negatively associated with diversity metrics). Perhaps these could be supplemental figures, and main figure space could be devoted to depicting patterns with key taxa discussed in the results.

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a valuable contribution to the sparse literature on the relationship between complementary feeding and infant microbiome.

General recommendations:

There is a general lack of scientific references and citations on cereal use in infancy in Spain/Europe

Are cereals important foods in Spain? Which grains? Are whole grain cereals common in Spain? What about other European countries?

Are infant cereals usually fortified in Spain? If so, how could fortification with iron and other minerals impact the microbiome?

The manuscript is missing discussion on the potential role of antinutrients like phytates in whole grain cereals for infants, -- what potential effects could whole grains have, if any, on infant health or the microbiome?

Please mention in title/abstract somewhere that this study is from Spain

Specific recommendations:

Line 61- “an increasing number of studies” Please cite which studies you are referring to

Some English language editing is needed throughout on verb/subject agreement, for example, Line 61- should be “have,” not “has”

Line 69 – “that smoothens the transition from exclusive milk feedings to a diverse and complete family diet” It is not clear what the authors mean by “smoothens the transition”

Table 1: Please give some basic information on the micronutrient content or mention in the text whether the cereals were fortified

Did the cereals need to be cooked by the parents first or were they ready-to-eat? Which kind of packaging was used--baby food jars or pouches or bags, etc? Were they cereal-milk porridges or just grain ingredients? 

Methods – Please give more information on the early nutrition of the infants (before complementary feeding). Were the infants breastfed or formula-fed? If they were formula fed, which kinds of formula? Was it infant formula with probiotics/HMOs? Were they already eating other complementary foods at the beginning of the trial? These factors could affect your results.

Line 225-226 the authors mention antibiotic use. Were any infants using pre/probiotics? Were they consuming yoghurt or other live culture food products?

Line 317- “Shannon, Pielou's, Simpson and inverse Simpson’s indices” Please use full names, it is not clear what these names mean, e.g. instead of Pielou’s, use “Pielou's evenness index”

Line 442- “Many national dietary guidelines support whole grain foods as the primary choice of grain products in the diet.” Please cite

Line 466—please cite

Conclusion – This section should focus more on the many interesting findings of this study relating to the microbiome. The sentences on 1) sustainable healthy diets and 2) acceptability of cereals, seem out of place, have not much to do with the content of the manuscript.

Reference 14- too old, there is recent literature on cereals in infancy from Europe

Reference 17 – Ok, but also rather refer to scientific literature on the addition of sugar to infant cereals

Author Response

Please see the attachment.

Author Response File: Author Response.pdf