First-Day-of-Life Rectal Swabs Fail To Represent Meconial Microbiota Composition and Underestimate the Presence of Antibiotic Resistance Genes

Abstract

The human gut microbiome plays a vital role in health and disease. In particular, the first days of life provide a unique window of opportunity for development and establishment of microbial community. Currently, stool samples are known to be the most widely used sampling approach for studying the gut microbiome. However, complicated sample acquisition at certain time points, challenges in transportation, and patient discomfort underline the need for development of alternative sampling approaches. One of the alternatives is rectal swabs, shown to be a reliable proxy for gut microbiome analysis when obtained from adults. Here, we compare the usability of rectal swabs and meconium paired samples collected from infants on the first days of life. Our results indicate that the two sampling approaches display significantly distinct patterns in microbial composition and alpha and beta diversity as well as detection of resistance genes. Moreover, the dissimilarity between the two collection methods was greater than the interindividual variation. Therefore, we conclude that rectal swabs are not a reliable proxy compared to stool samples for gut microbiome analysis when collected on the first days of a newborn's life. IMPORTANCE Currently, there are numerous suggestions on how to ease the notoriously complex and error-prone methodological setups to study the gut microbiota of newborns during the first days of life. Especially, meconium samples are regularly failing to yield meaningful data output and therefore have been suggested to be replaced by rectal swabs as done in adults as well. We find this development toward a simplified method to be producing dramatically erroneous results, skewing data interpretation away from the real aspects to be considered for neonatal health during the first days of life. We have put together our knowledge on this critical aspect with careful consideration and identified the failure of rectal swabs to be a replacement for sampling of meconium in term-born newborns.

Keywords: antibiotic resistance; gut microbiota; human microbiota; meconium; microbiome; newborns; rectal swabs; resistance genes.

FIG 1

Alpha diversity of the newborn’s gut microbiome differs greatly between the two sample collection approaches. Comparison of alpha diversity measurements assessed by calculating Shannon’s diversity index (A), detected number of species (B), and evenness index (C) for meconium samples and rectal swabs. Degree of significance was evaluated using the Wilcoxon rank sum test (*, P < 0.05; **, P < 0.01; ***, P < 0.001). (D to I) Depiction of alpha diversity measures for both sampling types on a sample-by-sample basis (D to F) as well as correlation between alpha diversity measurements for the two collection approaches (G to I) indicated lack of similarity between the sampling approaches.

FIG 2

Beta diversity analysis reveals distinct clustering for both sample collection methods. (A and B) Beta diversity was compared by computing principal-coordinate analysis (A) depicting distinct clusters for two sample groups (permutational multivariate analysis of variance using distance matrices, P < 0.01) and with assignment to variable color coding for the matched sample pairs (B). (C) Constrained correspondence analysis (CCA) proved the significant contribution of the sampling method (P < 0.01) and the nonsignificant impact of interindividual variation (P > 0.05).

FIG 3

Comparison of the taxonomic composition in meconium samples and rectal swabs indicates different microbial composition between the sampling strategies. Relative abundance of the 24 most prevalent taxa on the genus level is given in both sampling types. Taxa with higher abundance are listed at the top of the key.

FIG 4

Heatmap of taxa displaying significant differences between two sampling methods. Significance was calculated via pairwise Wilcoxon rank sum test (P < 0.05, with use of false-discovery rate adjustment of P values). Color coding depicts the Z-scores of average taxonomic abundance.

FIG 5

Heatmap depicting different detectabilities of antibiotic resistance genes in the sample types. Clustering was performed based on the Euclidean distance; light blue represents the absence of the resistance genes in the samples from one of the collection approaches, and dark blue represents presence in at least one of the samples of the group.