Sinus Microbial Communities in Patients With Chronic Rhinosinusitis Are Unpredictable Over Time

The role of microorganisms in chronic rhinosinusitis (CRS) remains unclear. High-quality cross-sectional studies have shown that CRS and healthy control bacterial communities exhibit subtle differences [1]. However, findings can vary between studies and most have not identified robust relationships between the microbiome and clinical variables [2, 3]. We hypothesize that the CRS bacterial community changes over time, which may help explain the inconsistencies in the literature. To test this hypothesis, we conducted a comprehensive longitudinal study to assess the stability of the bacterial community over one year. This study was approved by the New Zealand Health and Disability Ethics Committee (19/NTA/7/AM04). Adult patients with primary diffuse CRS were recruited from Auckland City Hospital. Our prospective longitudinal observational study design involved monthly visits to collect clinical data and samples. Visits were categorized into preoperative, recovery (<3 months after surgery), and postoperative (≥3 months after surgery) phases. The available medication data did not include the specific indications for antibiotic prescriptions. Samples from a previous study of four healthy participants over two years were included as controls [4]. Endoscopically guided middle meatal swabs were collected from each participant in the clinic, operating theater, or community using portable equipment and stored in tissue and cell fixative (RNAlater; Thermo Fisher Scientific, Waltham, MA, USA). The V3–V4 regions of the bacterial 16S rRNA gene were amplified for 2 × 300 base pair paired-end sequencing on the Illumina MiSeq platform (Illumina Inc., CA, USA). Amplicon sequence variants (ASVs) were assigned taxonomy using the Human Oral Microbiome Database (HOMD) version 15.22 [5]. The V1–V3 16S rRNA regions were amplified for droplet digital PCR on the QX200 Droplet Digital PCR System (Bio-Rad Laboratories Inc., Hercules, CA, USA). Statistical analyses were performed in R (version 4.0.5) [6]. Permutational multivariate analysis of variance (PERMANOVA) using Bray-Curtis distance matrices was used to assess factors influencing alpha diversity changes. Beta-diversity was determined by analyzing Bray-Curtis dissimilarities (BCD) using linear mixed-models and a Monte-Carlo permutation test. Controls were compared with preoperative CRS samples to exclude biases from surgery. Bonferroni–Holm and false discovery rate p value adjustments were applied. A p value of less than 0.05 was significant. All sequence data were deposited with National Center for Biotechnology Information (NCBI) under BioProject ID numbers PRJNA1184190 and PRJNA464064. A full description of the methods is available in File 1, Supporting Information. We compared 138 samples from 22 CRS participants with 48 samples from four healthy controls (Table S1). Sinonasal Outcome Test (SNOT-22) scores improved by 18.5 points following surgery (preoperative median 34, IQR, 20–52; postoperative median 15.5, IQR, 8.75–25.75; p < 0.001). In CRS, the most abundant ASVs were Corynebacterium accolens (14.7%), Corynebacterium pseudodiphtheriticum (12.6%), and Staphylococcus capitis (11.2%) (Figure 1). PERMANOVA showed that individual participants contributed 57.5% of the variability in bacterial communities over time (Table S2). Functional endoscopic sinus surgery (FESS), disease status, allergic rhinitis, and oral antibiotics contributed 6.7%, 4.8%, 3.8%, and 1.1%, respectively. Corynebacterium macginleyi relative abundance was associated with mild SNOT-22 scores (p < 0.0001), while Anaerococcus octavius was associated with severe scores (p = 0.008) (Figure S2). There were no significant differences in alpha diversity or beta diversity between preoperative CRS and control samples (Shannon p = 0.8; CRS BCD 0.53, control BCD 0.45, p = 0.3). No significant correlations were found between alpha diversity and acute exacerbations or polyp status in CRS. FESS with postoperative antibiotics decreased alpha diversity during recovery time points (Shannon beta −1.4, p < 0.001). FESS also decreased the relative abundances of C. accolens, C. pseudodiphtheriticum, and Dolosigranulum pigrum (p = 0.001) while increasing S. capitis (p = 0.001). Oral antibiotics decreased the relative abundances of S. aureus and Corynebacterium sp. (p < 0.001). Total bacterial load in preoperative CRS samples was 17 times higher than in controls (5120 median CRS 16S rRNA gene copies per swab, 359.2 median control 16S rRNA per swab, p < 0.001) (Figure 2). There were no significant differences in total bacterial load or Staphylococcus aureus load between pre- and postoperative CRS samples, although S. aureus load was higher during recovery (p ≤ 0.01) (Figure 2). Influenza A and COVID-19 were not detected. Rhinovirus was detected in seven samples but only two were associated with acute exacerbations (Table S3). These findings may have significant implications for how we approach CRS microbiome research, as cross-sectional data from a single time point may not accurately reflect the microbiome at other times. While several studies have documented changes in the CRS bacterial community, they have generally had shorter follow-up periods. One study found that surgery decreased alpha diversity and richness 12 weeks postoperatively [2]. Another study observed significant bacterial community changes 2 weeks after surgery, with a trend toward baseline by 6 weeks [3]. The most recent study reported a nonsignificant increase in alpha diversity 3 months postoperatively [7]. Some of these changes may reflect the temporary effects of FESS on mucosal inflammation, mucociliary clearance, and immune function. The limited impact of antibiotics on the bacterial community observed in this study, as well as previous studies [8, 9], suggests that their benefits may not primarily come from their antimicrobial properties. Study limitations include the small number of control participants, and the coronavirus disease 2019 (COVID-19) pandemic, which hindered some sample collections and led to fewer viral infections than usual [10]. In conclusion, our data suggest that CRS and healthy bacterial communities may be less stable than previously thought. These findings imply that designing and interpreting intervention studies aimed at altering the microbiome may be more challenging than previously perceived. We would like to acknowledge the Passe and Williams Foundation for their funding of the Academic Surgeon-Scientist Research Scholarship. We would like to thank Daniel Barnett and Danial Walsh from the Department of Statistics, University of Auckland, for their statistical analyses performed in this study. The authors declare no conflicts of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.