Chronic Rhinosinusitis: Potential Role of Microbial Dysbiosis and Recommendations for Sampling Sites

Abstract

Chronic rhinosinusitis (CRS) is an inflammatory condition that affects up to 12% of the human population in developed countries. Previous studies examining the potential role of the sinus bacterial microbiota within CRS infections have found inconsistent results, possibly because of inconsistencies in sampling strategies. The aim of this study was to determine whether the sinus microbiome is altered in CRS and additionally if the middle meatus is a suitable representative site for sampling the sinus microbiome. Swab samples were collected from 12 healthy controls and 21 CRS patients, including all eight sinuses for CRS patients and between one and five sinuses for control subjects. The left and right middle meatus and nostril swabs were also collected. Significant differences in the sinus microbiomes between CRS and control samples were revealed using high-throughput 16S rRNA gene sequencing. The genus Escherichia was over-represented in CRS sinuses, and associations between control patients and Corynebacterium and Dolosigranulum were also identified. Comparisons of the middle meatuses between groups did not reflect these differences, and the abundance of the genus Escherichia was significantly lower at this location. Additionally, intra-patient variation was lower between sinuses than between sinus and middle meatus, which together with the above results suggests that the middle meatus is not an effective representative sampling site.

Keywords: 16S rRNA gene sequencing; chronic rhinosinusitis; microbiome; middle meatus; sinus.

Figure 1

Boxplot of Shannon diversity. (A) Diversity in sinus cavities from CRS subjects separated by the absence (CRSsNP) or presence (CRSwNP) of nasal polyps and control subjects. Significant differences indicated by ^* (ANOVA, p < 0.05).

Figure 2

PCoA of weighted Unifrac distances. Samples from sinus cavities only are included, and ellipses represent 95% confidence intervals.

Figure 3

Average relative abundance of phyla in CRS and control subjects. Values are averages across all sinus samples per patient. Percentages were calculated from rarefied data. Phyla that were significantly different in relative abundance between CRS and controls are marked with ^* (Kruskal-Wallis, FDR corrected p < 0.05).

Figure 4

Average relative abundance of OTUs per patient colored by genus. Only sinus samples were included, and OTUs with average relative abundance of <1% were excluded. Percentages were calculated from rarefied data. Genera that were significantly different in relative abundance between CRS and controls are marked with ^* (Kruskal-Wallis, FDR corrected p < 0.05).

Figure 5

Network of OTUs associated with disease status, and strong significant correlations to other OTUs. Networks were calculated using extended local similarity analysis and plotted using the Cytoscape3 software package. Nodes are colored by genus taxonomic assignment, and metadata category nodes are colored red. Glow behind nodes indicates the relationship to disease status (blue, positive correlation; red, negative). Black lines indicate positive correlations, and the length indicates the strength (short, strong; long, weak). Red lines indicate negative correlations and the length indicates the strength in the opposite direction to black (long, strong; short, weak).

Figure 6

SNOT-22 vs. relative abundance for (A) one Corynebacterium OTU with a negative correlation, and (B) the genus Escherichia with a positive correlation. Plots are overlaid with a linear regression line and shading indicating 95% confidence interval. Individual patient samples are indicated with color. Rarefied data were used to generate the plot.

Figure 7

Differences between middle meatus and sinus microbial communities. (A) Intra-patient distances between middle meatus and sinus (MMvS) and sinus and sinus (SvS) samples for weighted Unifrac distances. Significant difference is indicated with ^* (ANOVA, p < 0.05). (B) Average relative abundance of the genus Escherichia in middle meatus samples (top panels), and sinus samples (bottom panels).

Figure 8

Intra vs. inter-individual weighted Unifrac distances in CRS and control subjects. Inter-individual distances were significantly higher than intra-individual distances, indicated with ^*(ANOVA, p < 0.05). A trend of higher intra-individual distances in CRS subjects was observed, but only approached significance (ANOVA, p = 0.06).