Perinatal factors affect the gut microbiota up to four years after birth

Perinatal factors impact gut microbiota development in early life, however, little is known on the effects of these factors on microbes in later life. Here we sequence DNA from faecal samples of children over the first four years and reveal a perpetual evolution of the gut microbiota during this period. The significant impact of gestational age at birth and delivery mode on gut microbiota progression is evident in the first four years of life, while no measurable effects of antibiotics are found in the first year. Microbiota profiles are also characteristic in children dependant on gestational age and maturity. Full term delivery is characterised by Bacteroides (year one), Parabacteroides (year two) and Christensenellaceae (year four). Preterm delivery is characterised by Lactobacillus (year one), Streptococcus (year two) and Carnobacterium (year four). This study reveals that the gut retains distinct microbial profiles of perinatal factors up to four years of age.


2.
Figures are of very low quality and some are illegible.

3.
It is unclear how the results described in lines 79-81 are reflected in Figure 1.

4.
Line 403: State version of QIIME Reviewer #2 (Remarks to the Author): In the paper "Microbiome Memory: Perinatal factors continue to affect the gut microbiome four years after birth" Fouhy et al describe the influence of perinatal factors on the development of the microbiome during the first years of life. The paper is written beautifully however the authors should emphasize much more it's innovation as there are several paper looking at time series of babies. My major concern with the work presented is that it seems that most of the data is not significant after corrections which I think is a problem.
Some more specific comments: • Figure 3A is unreadable • Line 115 -The authors should show the data as it is of interest.
• Line 132/figure 6 -what statistics were done • Figure 6 is unreadable • Throughout the paper in each analysis the authors take into account a different number of the most abundant OTUs. I do not think this is correct as some of the more interesting OTUs might be of low abundance. In any case the authors need to explain the different numbers in each analysis. For example, line 141 -50 most abundant, line 165 -300 most abundant, line 132 -200 most abundant etc. It appears that if the authors were to use a different cutoff than no OTUs would pass FDR.

•
It is unclear why the authors mention differences and then state that these differences were no longer apparent after adjustment for condition or FDR. All results that are not significant after adjusting or FDR should be removed. For example, lines 185-193, lines 172-180, line 195-203  This study include 112 individuals with gut microbiota samples taken at 1, 2, and 4 years. 16S rRNA sequencing was used to process the samples. Numerous early life exposure were examined, including gestational age at birth, delivery mode, and antibiotic usage, to assess the association with the gut microbiota alpha diversity and composition at the examined sampling times. They found no evidence that antibiotic exposure in the first year significantly shaped the later gut microbiota, but found that gestational age at birth had sustained effects, as did delivery mode.

General comments:
The writing style is somewhat technical and difficult to follow. This article could not be easily understood by someone outside of the field of microbiome research, and the paper sometimes reads like a summary of results from all of the possible analyses in a microbiome software package.

Response:
We have taken on board this feedback and rephrased the manuscript. We have included improved descriptions and explanations of the statistical analysis performed (lines 389-404) and we have included plain language summaries of each stats package run throughout the manuscript e.g. (105-106, 114-116, 130-131, 172-175) and the reasons for choosing the approaches used. We have reduced the technical language used and think the revised manuscript is easier to interpret.
The discussion does not clearly convince me of the importance or impact of these results.

Response:
We have rewritten the discussion to expand on the importance and novelty of the results. We have included reference to relevant publications and how our research adds to existing findings (e.g. 295-296, 320-322, 300-302).
Furthermore, some of the details of the cohort are not clearly described, such as how many of children have 1, 2 and 3 gut microbiota samples, the details of the collected exposure data, etc. A table with cohort characteristics for each time should be included. The paper suggests that only 52 have more than one sample, but no further detail is given. It is difficult to assess the appropriateness of all of the performed analyses without better understanding the longitudinal nature of the data. There is a huge amount of temporal and inter-individual variation in early life in the gut microbiota (particularly with a wide range of perinatal exposures).
In the paper "Microbiome Memory: Perinatal factors continue to affect the gut microbiome four years after birth" Fouhy et al describe the influence of perinatal factors on the development of the microbiome during the first years of life. The paper is written beautifully however the authors should emphasize much more it's innovation as there are several paper looking at time series of babies.
Response: Thank you for your compliments regarding the writing style of our manuscript. We have rephrased the revised manuscript to further highlight the innovation (line 232-324, 340-341, 343-346). This study investigated whether perinatal factors including delivery mode, gestational age at birth and feeding regime would result in distinct microbial profiles extending to four years of age.
My major concern with the work presented is that it seems that most of the data is not significant after corrections which I think is a problem.

Response:
In response to concerns from Reviewers 1 & 2 we have rerun the analysis using ANCOM and rewritten the associated results (line 394-397, 172-179, 202-225). This statistical analysis is a more suitable and robust statistical analysis method for gut microbiota data and has been shown to have the most sensitive FDR corrections. All of the p values in the text are the corrected p values (lines 205, 206,176-179).

Some more specific comments:
• Figure  etc. It appears that if the authors were to use a different cutoff than no OTUs would pass FDR.
We have also included this information in all of the revised figure legends. We tested the data using much larger numbers of up to 20,000 taxa and this did not affect the results achieved. Therefore, we have now included 1000 taxa in all analysis. The exception is for the heatmap, as using greater than 100 taxa made the image too crowded and illegible (lines 138, 548). However, the pattern of clustering remained the same when we tested this analysis using 1000 taxa.
• It is unclear why the authors mention differences and then state that these differences were no longer apparent after adjustment for condition or FDR. All results that are not significant after adjusting or FDR should be removed. For example, lines 185-193, lines 172-180, line 195-203 etc. Response: In response to concerns from Reviewers 1 & 2 we have rerun the analysis using ANCOM and rewritten the associated results (line 394-397, 172-179, 202-225). This statistical analysis is a more suitable and robust statistical analysis method for gut microbiota data and has been shown to have the most sensitive FDR corrections. All of the p values in the text are the corrected p values (lines 205, 206,176-179).
• Line 297 change was to were. This manuscript reads much more clearly after the edits, and the clarification of the details of the cohort and methods, as well as the changes in the statistical analyses, improve the strength of the manuscript.

Response
A few additional comments: Lines 59-60: I am not sure that there is strong evidence that gestational age is the "primary" driver of early life gut microbiota. There is evidence that it is important, but I believe that this finding that gestational age was the main driver of the gut microbiota development was among very low birth weight infants, and the uniqueness of this cohort might be very relevant to their conclusions. I think that this results should be restated and clarified.
Line 51: This sentence is unclear. Are only the FT infants born vaginally or by CS? Are all PT born via CS?
There is inconsistency in the number of significant digits reported for p-values.
The statistical methods do not describe all analyses performed. The methods section should be thorough enough for someone else to repeat the analyses. For example, the alpha and beta diversity analyses are not adequately described. The results reported in lines 120-128 seem to suggest that Adonis was used incorrectly. It is not meant for repeated measures data (I believe that nested.anova.dbrda allows for this), yet there is only one p-value reported across the years, suggesting that all data were included.
Line 151: What are optional study groups?
Some of the sections seem to contain material that belongs in other sections, eg. Line 153 is a method not a result.
Line 246: Is this an analysis of age of the child (1 year, 2 year, etc) or gestational age? This is unclear and not clarified by the title of the supplemental table (which did not convert clearly to PDF).
Reviewer #2 (Remarks to the Author): I thank the reviewers for addressing my concerns. However I still feel that the innovative aspect of this paper is not highlighted enough.
We thank both reviewer's for their comments and feedback. We have revised our manuscript accordingly and address each of their comments in detail below.

Reviewer #1 (Remarks to the Author):
This manuscript reads much more clearly after the edits, and the clarification of the details of the cohort and methods, as well as the changes in the statistical analyses, improve the strength of the manuscript.
A few additional comments: Reviewer 1: Lines 59-60: I am not sure that there is strong evidence that gestational age is the "primary" driver of early life gut microbiota. There is evidence that it is important, but I believe that this finding that gestational age was the main driver of the gut microbiota development was among very low birth weight infants, and the uniqueness of this cohort might be very relevant to their conclusions. I think that these results should be restated and clarified.
Response: Infants can be born too early (<35 weeks), too small (<2500g) or both, therefore not all low birth weight infants are preterm. In the absence of accurate gestational age information (e.g. by first trimester ultrasound) previous studies have used birth weight as a surrogate marker of prematurity. However, in our opinion gestational age is a more accurate measure, as not all low birth weight infants are preterm (e.g. in the case of placental insufficiency and intrauterine growth restriction). For example, in our study 13% of preterm born individuals were > 2500g. The impact of birth weight on gut microbiota has now been added to lines 58-70 and we have also highlighted the impact of other factors associated with premature delivery on gut microbiota (e.g. CS delivery). Our study found that prematurity exerts a persistent effect on the microbiome. Although we cannot predict the long term health implications of the altered microbiome in prematurity, it does give impetus for large scale studies to examine the relationship between developmental delay and the microbiome in premature infants and determining whether the persistence of these microbial signatures is related to developmental consequences.  16,17,18 . Indeed, many of these PT studies include very low birth weight (VLBW) infants with extended hospitalisation care, thereby exposing the juvenile microbiome to surfaces in the NICU which have previously been shown to influence gut microbiota colonisation 19 . Birth weight is therefore an underlying factor which plays a role in microbiota progression as VLBW infants also possess an immature immune system which influences microbe-gut interactions 20 ",