After approval from the ethics committee of the Medical University of Vienna (EK No. 2027/2019), subject recruitment and clinical study were conducted at the Department of Otorhinolaryngology of the Medical University of Vienna. Written informed consent was obtained from each patient before study inclusion according to the Declaration of Helsinki. A total of 80 patients were recruited (20 per group for CRSwNP, CRSsNP, N-ERD, and DC).

CRS was diagnosed based upon the EPOS 2020 criteria (4). Furthermore, the diagnosis of N-ERD was defined according to the European Academy of Allergy and Clinical Immunology (EAACI) position paper as the presence of nasal polyps and asthma, in addition to developing respiratory symptoms upon the ingestion of aspirin or NSAIDs (6). In detail, 11 patients underwent a positive oral challenge test with NSAID. The remaining 9 patients had in the past either suffered from an anaphylactic shock (n=2), attended an emergency unit (n=3) or had reportedly repeated respiratory and nasal symptoms specifically after ingestion of NSAID. Exclusion criteria for participation in the study included age <18, pregnancy (standard urine pregnancy test was performed at the study visit), the use of nasal or systemic corticosteroids or immunosuppressants two weeks prior to their visit, use of biologicals, the presence of cystic fibrosis, and severe anatomic variations of the nasal cavity not allowing access with the nasal swabs.

During their visit to the outpatient clinic and after signing the informed consent, a medical history with regard to allergies, asthma, and nasal diseases, including previous surgeries and therapy, was recorded using a questionnaire. To assess the disease-specific and general health-related quality of life (QoL) in our study cohort, we used the sino-nasal outcome test-22 (SNOT-22), the EuroQol (EQ)-5D-3L, the Patient Health Questionnaire (PHQ)-2, and the asthma control test (ACT). Furthermore, scoring of polyp size was performed during routine nasal endoscopy by an otorhinolaryngologist. The total polyp score (TPS) system was used to grade nasal polyp size by nasal endoscopy. Both sides of the nasal cavity were separately assessed and scored, resulting in scores of 0–4 (0 = no polyps, 1 = small polyps in the middle meatus not reaching below the inferior border of the middle turbinate, 2 = polyps reaching below the lower border of the middle turbinate, 3 = large polyps reaching the lower border of the inferior turbinate or polyps medial to the middle turbinate, 4 = large polyps causing complete obstruction of the inferior nasal cavity) (20). The sum of both nostril scores was considered as the TPS. Olfactory testing was performed using the Sniffin Sticks Threshold (T), Discrimination (D) and Identification (I) test (Burghart, Holm, Germany). The sum of the three subtests resulted in the TDI score (21). For allergy assessment, a standard skin prick test was performed. Additionally, allergen-specific IgE levels and total IgE were determined in serum using the multiplex ALEX2 microarray (Allergy Explorer, MacroArray Diagnostics GmbH, Vienna, Austria). Patients’ characteristics and clinical measurements are displayed in Table 1 and Table E1 .

First, nasosorptions were used for the collection of nasal secretions from the inferior nasal turbinate (Nasosorption FX-I, Hunt Developments (UK) Limited, Midhurst, West Sussex, United Kingdom) (22). The devices were inserted into the nasal cavity under visualization and placed along the lateral wall of the inferior turbinate. The patient held the device in place by pressing onto the external aspects of the alar and nasal cartilage for one minute. Thereafter the device was removed, placed on ice and immediately transported to the laboratory. The collected nasal fluid was eluted using 300 μL of phosphate-buffered saline (PBS) containing 0.05% Tween 20 and 1% bovine serum albumin (BSA) per nasosorption. Samples were immediately frozen at -80°C until further analysis for cytokines.

Thereafter, swabs optimized for the collection of microbiome specimens were applied (CLASSIQSwabs, Copan Diagnostics Inc. Murietta, CA, USA) to the anterior naris and middle meatus of each nostril. In addition, air controls were performed on each sampling day. Samples were immediately frozen at -80°C for later analysis by 16S rRNA gene amplicon sequencing.

Mediator concentrations in nasal secretions were measured using electrochemiluminescence technology through the use of the mesoscale discovery platform (MSD, Rockville, MA, USA). The following cytokines were analyzed: Eotaxin, Eotaxin 3, G-CSF, GM-CSF, IFN-γ, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, IL-16, IL-17A, IL-1α, IL-1RA, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-21, IL-22, IL-17E/IL-25, IL-27, IL-33, CCL17, TNF-α, TNF-β, TSLP, and VEGF-A. Custom cytokine plate kits were purchased from MSD and analysed according to the manufacturers protocol: Briefly, first linker-antibody solutions were prepared according to the protocol and then used to coat the U-Plex plate (50 μl per well), plates were sealed and left shaking overnight at 2°C–8°C. To generate the standard curve, the calibrator was diluted 1/10 in the provided metabolic assay working solution, and 8 standards were subsequently created with 1 in 4 dilution steps. After standard preparation was complete, the coated plates were washed 3 times with 150 μl of PBS-T (PBS 0.05% Tween 20) in each well. To each well, 50µL sample (pooled from right and left nostril of individual patients) or standard were added, the plate was sealed and incubated with agitation for 2 h at room temperature. Thereafter, plates were washed 3 times with PBS-T, and 50 μl of the provided detection antibody solution were added to each well, followed by another incubation period of 1 h while shaking. The plate was again washed 3 times with PBS-T, and 150 μl of the kit supplied MSD GOLD read buffer was added to each well. The plates were then analyzed using the MESO SECTOR S 600 plate reader (MSD), and cytokine concentrations were calculated using the DISCOVERY WORKBENCH software (MSD). Missing concentration values resulting from readings below or above the detection limit were set to lower or upper detection limit ( Table E2 ) of the cytokine respectively.

Statistical analyses of the clinical data were done using GraphPad Prism 8.4.1 (GraphPad Software, Inc., La Jolla, San Diego, CA, USA) after consultation with the Center for Medical Statistics, Informatics and Intelligent Systems, Medical University of Vienna. To normalize cytokine values for subsequent analysis, variables were transformed by log transformation. To test for statistically significant differences with regards to clinical parameters (i.e. TPS, SNOT-22) and cytokine concentrations between the 4 different groups, i.e. DC, CRSsNP, CRSwNP and N-ERD, the Kruskal-Wallis test was applied due to not normally distributed values (For results of Kruskal-Wallis tests for all figures, please refer to Table E3 ). This was followed by the Dunn’s test for multiple comparison. To report TPS, SNOT-22 and results of the quality of life questionnaires, descriptive analyses were used and described by median and range values. Statistical significance was defined as a P value of less than 0.05 and is indicated in the respective figures.

16S rRNA gene sequencing and raw data processing was performed at the Joint Microbiome Facility of the Medical University of Vienna and the University of Vienna (project ID JMF-2102-04) as previously described (23). Samples were sequenced in two batches (runs). Amplicon sequence variants (ASVs) were inferred using the DADA2 R package v1.20 (24) applying the recommended workflow (25). FASTQ reads 1 and 2 were trimmed at 230 nt and 210 nt with allowed expected errors of 4 and 6, respectively. ASV sequences were subsequently classified using SINA version 1.6.1 (26) and the SILVA database SSU Ref NR 99 release 138.1 (27) using default parameters.

The resulting ASVs were stringently filtered to retain only trusted results. We only kept ASVs that were detected in at least 20% of non-negative control samples in both amplicon sequencing runs where the data was generated in; other ASVs were considered batch-effect contaminants. We also removed ASV_ku3_j7x (Homo sapiens mitochondrion) and ASV_6em_2nf, ASV_kip_nj9, ASV_paz_uyl based on manual inspection, as these were likely swab/reagent contaminant due to also being found in more than 4 negative controls. After merging the data from left and right nostril samples, all samples that had <1000 reads, as well as all samples from MN062, MN072, MN073, MN074 and MN075 (as these patients had a very high yield in their corresponding sampling negative control), were omitted from downstream analyses.

Following filtering, downstream analyses were performed using R v4.2 and Bioconductor v3.15 packages TreeSummarizedExperiment v2.4 (28), mia v1.4 (https://github.com/microbiome/mia), vegan v2.6.2 (https://CRAN.R-project.org/package=vegan), phyloseq v1.40 (29), microbiome v1.18 (http://microbiome.github.io), microViz v0.9.1 (30), DESeq2 v1.34 (31), ALDEx2 v1.28 (32). Alpha diversity was calculated using vegan and mia on rarified data (1007 read pairs). Beta diversity was calculated by performing a PCA on centered log ratio-transformed data using microViz. The difference in per-group centroids was tested with a PERMANOVA on Aitchison distance using vegan and microViz ( Table E4 ). For PERMANOVA analysis, samples were grouped by disease group, presence and absence of allergy or asthma. Interactive effects between disease group and patient age, duration of CRS and number of polypectomy surgeries were tested, too. Pairwise differential abundance testing was performed using DESeq2 with alpha=0.05 and otherwise default parameters after adding a pseudocount of 1 to the data. Correlation between amplicon sequence variant (ASV) centered log-ratio transformed counts and environmental variables was calculated using ALDEx2’s correlation test using Pearson correlation with FDR multiple testing correction. Data accessibility: Amplicons datasets generated under JMF Project ID JMF-2102-04 and used in this study are deposited under the BioProject ID PRJNA907202.

A total of 80 patients were included in this study (n=20 per group of DC, N-ERD, CRSwNP, CRSsNP). Patients suffering from N-ERD or CRSwNP had a mean age of 46.8 ± 10.0 and 49.4 ± 11.9 and were older than those suffering from CRSsNP (mean ± standard deviation 33.1 ± 10.0) or disease controls (mean ± standard deviation 25.8 ± 3.8) ( Table 1 , p-values: Table E3 for Kruskal-Wallis and Table E4 for PERMANOVA results). Comorbidities included respiratory allergies and asthma. The latter was more prevalent in patients with N-ERD (100%) and CRSwNP (80%). Allergic sensitization to respiratory allergens was similarly frequent among the 4 groups (Respiratory allergic subjects per group: DC n=9, CRSsNP n=7, CRSwNP n=8, N-ERD n=9). PERMANOVA analysis revealed that age, presence of allergy or asthma, number of previous surgeries as well as duration of CRS did not significantly influence microbiome composition ( Table E4 ). TPS score tended to be higher in N-ERD (median 5; range 0-8) as compared to CRSwNP patients (median 4; range 0-8), but did not reach significance ( Figure 1A ). Both N-ERD (median 12.5; range 0-34.8) and CRSwNP (median 14; range 7-24.8) had significantly reduced smell perception as measured by TDI scores in comparison to CRSsNP (median 32; range 13-38) and disease controls (median 32.5; range 14.5-37.5), whilst no difference in TDI was observed between CRSsNP and DC ( Figure 1B ).

Regarding disease-specific and general health-related QoL, only the SNOT-22 showed significant differences between patients with CRS and DC (median 48; range 19-103 in N-ERD, 41; 9-76 in CRSwNP, 46; 11-84 in CRSsNP vs. 0; 0-11 in DC; P ≤ 0.0001) ( Figure 1C ). This difference was evident across all 4 subdomain analyses (nose, ear/face, sleep, emotion) with the nose domain yielding the highest scores ( Figure E1 ). No major differences were observed in the EQ-5D-3L ( Figure 1D ), PHQ-2 ( Figure 1E ), and ACT questionnaires ( Figure 1F ). However, it needs to be noted that only one patient in the DC and three in the CRSsNP group suffered from asthma and completed the ACT. Therefore, the latter results need to be interpreted with caution.

Levels of a panel of 33 different cytokines were measured using the MSD Platform. We observed significantly elevated levels of the type 2 response-associated cytokines IL-5, IL-9 and Eotaxin-3 in the nasal secretions of patients with NP (N-ERD and CRSwNP) compared to patients without NP (CRSsNP and DC) (please refer to Figures 2A–C for p-values). Eotaxin was significantly elevated in CRSwNP patients only ( Figure 2D ), and CCL17 was significantly higher in CRSwNP and N-ERD patients than in CRSsNP and DC ( Figure 2E ). DC showed significantly higher levels of IL-1alpha than all other patient groups ( Figure 2F ). In contrast, the inflammatory cytokine IL-6 was mainly significantly elevated in CRSwNP patients ( Figure 2G ). All other cytokines did not show major differences between the different disease entities and disease controls ( Figure E2 ).

16S rRNA gene amplicon sequencing was performed in samples from the anterior naris (AN) and middle meatus (MM) in all patients. As low microbial biomass samples are more prone to be affected by exogenous contaminants which compromise data quality (12), the dataset was first stringently decontaminated, removing all ASVs detected as prevalent in PCR negative controls, DNA extraction blanks and air controls (see details in methods), and samples containing less than 1000 reads after decontamination were removed from the downstream analysis (available microbiome data: DC: n_AN=19, n_MM= 11; CRSsNP: n_AN=13, n_MM= 8; CRSwNP: n_AN=13, n_MM= 8; N-ERD: n_AN=14, n_MM=7). To assess intra-community diversity (alpha diversity), the Chao1 index focusing on total species richness, and the Shannon index assessing evenness in addition to richness, were calculated. For both indices, no significant differences between the patient groups and disease controls were noted ( Figure 3A ). To assess intersample variability, beta diversity was calculated and also showed no significant difference between the four groups, as depicted in multidimensional scaling plots in Figure 3B . ASVs affiliated with the genus Corynebacteria exhibited the highest relative abundance in both middle meatus and AN samples of all patients ( Figure 4A ). Staphylococci, Peptoniphilus, and Finegoldia-related ASVs showed a trend towards higher prevalence in AN of patients than in the MM. Lower relative abundance was observed for Dolosigranulum ASVs in AN samples. Pairwise differential abundance testing showed a significantly lower relative abundance of Corynebacteria in the AN of patients suffering from N-ERD and a higher relative abundance of Staphylococci in the MM of N-ERD patients as compared to other patients or disease controls ( Figure 4B , p-values indicated in the figure). The genus Dolosigranulum was relatively less prevalent and the genus Lawsonella was relatively more prevalent in the nose of patients suffering from nasal polyps than in patients without polyps and disease controls. In the latter group, a higher relative abundance of Finegoldia was observed.

Last, we aimed to delineate associations between cytokine responses and clinical characteristics of patients with the relative abundance of different bacteria taxa. After adjustment for multiple testing, Staphylococcus-related ASV (Staphylococcus argentus/aureus/equorum/phage/schweitzeri/simiae) and IL-5 showed a moderate strong correlation ( Figure 5A , Pearson r=0.55). At the same time, the relative abundance of an ASV classified as Corynebacterium accolens and Eotaxin-3 were negatively correlated in both MM and AN, but only showed a strong moderate correlation in MM ( Figure 5B , Pearson r=-0.54).