CF patients and healthy controls were prospectively included between 01/2013 and 02/2014. The recruitment of pwCF took place during regular outpatient visits to the Jena University CF Center. Healthy controls were recruited by a student from the CF center and other staff of the Jena University Hospital. CF diagnosis of patients relied on two positive sweat tests and/or identification of two disease-causing mutations in the CFTR gene locus.

Acute pulmonary exacerbation was defined according to Fuchs criteria in its modified version (35). A patient was considered to have an exacerbation when 4/12 of the following signs or symptoms were met: change in sputum, hemoptysis, increased cough, increased dyspnea, malaise/fatigue or lethargy, body temperature >38°C, weight loss/loss of appetite, sinus pain, altered sino-nasal secretions, change in physical examination of the chest, and decrease of 10% or more in forced expiratory volume in 1 s (FEV₁) compared to previous test results and radiographic signs indicating pulmonary infection.

Inclusion criterion for stable/healthy phase was the absence of symptoms for at least 14 days prior to nasal lavage collection. Exclusion criterion for the control cohort was the diagnosis of CF, primary ciliary dyskinesia, or immunodeficiency. Subjects with tympanic membrane perforation were excluded from both cohorts. Treatment with inhaled or oral antibiotics, including azithromycin, was ascertained but not regarded as exclusion criterion. In contrast, current administration of intravenous antibiotics was an exclusion criterion for “stable phases.” Further medical treatment was documented including nasally or pulmonary applied topical steroids.

All subjects below the age of 18 years were included in the pediatric cohort and subjects aged at least 18 years in the adult cohort. For evaluation, subjects turning 18 years old during the study remained in the pediatric cohort.

Written informed consent was obtained from each patient or his/her legal guardian, and the study was approved by the local ethics committee (technical opinion number 3608-11/12) in accordance with the 1964 Helsinki Declaration. The study was registered on Clinical trials.gov (NCT00803881).

NLF was obtained according to the procedure described in previous publications (36). In brief, using a syringe, 10 ml of sterile isotonic saline was applied to each nostril. Patients were told to hold their head in a slightly backward position for 10 s. During the procedure, the soft palate was temporarily closed forming a “K”-sound, preventing fluid from dripping into the oropharynx. Thereafter, while leaning slightly forward during exhalation, NLF was rinsed into a sterile beaker. If conventional nasal lavage (NL) was not possible because of age and/or compliance, we applied compressor-assisted NL using a large drop-nebulizing device with 10 ml isotonic saline (Rhinoclear^®). NLF was collected, and samples were treated as mentioned above. Samples were either taken during regular outpatient clinic visits where most aliquots were directly frozen at −75°C or at home by the patients/families. When samples were taken from CF patients at home, one aliquot was separated and directly sent to the laboratory for conventional microbiological analysis. The remaining samples were immediately frozen at home at −20°C and later transferred on ice to our clinic for assessment of inflammatory mediators. Further storage for all samples occurred at −75°C until analysis ( Figure 1 ).

Bacterial pathogen detection in pwCF was performed via conventional microbiology using native NLF samples in the Jena University Hospital Department of Microbiology, following the current standards. The extent of bacterial colonization, i.e. intermittent or chronic, was defined according to Lee criteria (37): chronic infection was defined as >50% of positive microbiological examinations during the preceding 12 months and intermittent infection as <50% of positive microbiological examinations during the preceding 12 months (at least three samples taken/year). However, both groups were later combined due to their small sample size. Modified Lee criteria were also applied on every bacterium found in the UAW. As only 1/49 pwCF was found positive with methicillin-resistant S. aureus (MRSA), no distinction was made between colonization with MRSA and methicillin-sensitive S. aureus (MSSA). Similarly, UAW colonizations of non-mucoid and mucoid P. aeruginosa were combined. During this study, P. aeruginosa colonization was detected in 16/49 patients. Additionally, four patients carried P. aeruginosa intermittently during the preceding year, according to retrospective chart review.

The physiological flora of UAW was defined according to (38, 39). Bacteria causing exacerbation and ARI were defined according to Goss et al., Zemanick et al., and others (40–42). For microbiological analysis, specimens were processed at the Jena University Hospital microbiology laboratory according to the German quality assurance guidelines for CF microbiology (43, 44).

Nineteen bacterial and viral pathogens known to cause acute airway infections/exacerbations were detected using a real-time multiplex PCR (m-RT-PCR) instrument (Roche, Mannheim, Germany) at the University Hospital of Schleswig-Holstein, Kiel, Germany. Viral pathogens targeted by PCR were adenovirus, enterovirus, influenza virus A (including H1N1) and B, parainfluenza viruses 1–4, respiratory syncytial virus (RSV), human rhinovirus (hRV), metapneumovirus (hMPV), coronaviruses 229E and OC43, and human bocavirus. Bacterial pathogens targeted by PCR were Chlamydia pneumoniae, Legionella pneumophila, Bordetella pertussis, Bordetella parapertussis, and Mycoplasma pneumoniae (45).

In six samples (one stable phase sample and five exacerbation samples, respectively) in which conventional microbiology could not be performed due to an insufficient amount of material, multiplex PCR analysis of viral- and pneumonia-causing pathogens was conducted using the Unyvero Hospitalized Pneumonia (HPN) Panel (Curetis GmbH, Holzgerlingen, Germany). Samples were thawed at room temperature and treated according to manufacturer guidelines.

NLFs were thawed and centrifuged at 1,000 rpm for 10 min. Cytokine levels were measured via bead-based multiplex assay using the Milliplex MAP-Kits^® (Human High Sensitivity HSTC-MAG-28K, Human MMP Panel 2 HMMP2MAG-55K, Human TIMP Panel 1 HTMP1MAG-54 K, Merck Millipore, Darmstadt, Germany). Fluorescence analysis was conducted using a fluorimeter (Bio-Plex^®200 System, Bio-Rad, CA, USA). Concentrations of IL-1β, IL-6, IL-8, matrix metalloproteinase (MMP)-9, and tissue inhibitor of metalloproteinase (TIMP)-1 were determined according to instructions provided by the manufacturer. As specified by the manufacturer, minimum detection limits were 0.12 pg/ml (IL-1β), 0.13 pg/ml (IL-6), 0.12 pg/ml (IL-8), 2.0 pg/ml (MMP-9), and 4.0 pg/ml (TIMP-1). Samples below the minimum detection limit were replaced by their respective detection limit minus 0.1 pg/ml. Samples above 1.5 times of the maximum detection limit were diluted (dilution factors, 1:5 to 1:10) with isotonic (for IL-1ß, IL-6, and IL-8) or buffered saline (for MMP-9 and TIMP-1) before rerun. In 21/214 samples, further dilution did not lead to decreased values. In these cases, the maximum value in the series of measurements of the respective parameter plus 0.1 pg/ml was assumed. Neutrophil elastase (NE) was determined using Human PMN Elastase ELISA (DEH3311, Demeditec Diagnostics GmbH, Kiel, Germany). The detection limit specified by the manufacturer was 0.2 ng/ml. All samples were used undiluted, and measurements were taken out in duplicates. Spectroscopy analyses were performed using FLUOstar Galaxy Spectrometer (BMG LABTECH GmbH, Offenburg, Germany).

Statistical analyses were performed with SPSS 27 (IBM, Ehningen, Germany), MS Excel (Redmont, USA), and GraphPad Prism 8 (La Jolla, USA). Sample sizes for the patient and healthy control cohorts were calculated with preliminary data of IL-1β, IL-6, IL-8, NE, TIMP-1, and MMP-9 in NL of 12 patients and 7 healthy controls at exacerbation. Given that such data were not normally distributed, sample size calculations were based on medians of such inflammatory parameters, according to (46) and assuming standard deviations equal to 0.5-fold of the median of each respective inflammatory parameter. Of all inflammatory parameters, NE yielded the largest estimated sample size (n=35) to attain a minimum power level of 0.8. Therefore, considering a dropout rate of approximately 10%, i.e., approximately n=3, we arrived at an estimated minimal sample size of n=38. Descriptive statistics of demographic data was reported as means and their standard deviations. Comparisons between percentages of individuals infected with bacterial, fungal, and viral pathogens were conducted using Pearson chi-squared tests. Results from comparisons of inflammatory markers are reported as medians and interquartile ranges (IQRs). Statistical significance was determined using mixed analysis of variance (ANOVA) on ranks for comparisons of inflammatory markers in CF patients and healthy controls at stable and exacerbation phases, and children and adult subgroups. When the normality assumption for mixed ANOVA was not met (as in IL-1β and IL-8 cases), log10 transformations were used. Statistical analyses of inflammatory markers between CF subgroups (CF patients receiving and not receiving antibiotics, with P. aeruginosa colonization, and with viral infections) were performed via Mann–Whitney U-tests. Odds ratio calculations were performed using the generalized estimating equations method for repeated measures and assuming an unstructured correlation matrix. p-values <0.05 were considered statistically significant.

During the study period, a total of 214 NLF samples, obtained from 49 pwCF (126 NLF samples) and 38 healthy controls (88 NLF samples), were collected at five time points (see Supplementary Table S7 ) with a mean of 2.5 NL per patient in the CF cohort (range, 2–5) and a mean of 2.2 NL in controls (range, 2–5). Sex ratio in both groups was non-significantly imbalanced in favor of female individuals in the CF cohort (22m, 27f). This imbalance was even more pronounced in controls (12m, 26f). Mean age at study entry was similar in both cohorts.

Among pwCF, 21/49 (43%) were homozygous and 26/49 (53%) were heterozygous for the most common mutation F508del. A total of 2/49 (4%) patients had rare non-F508del CFTR mutations (R347P/2183AA->G and G542X/D110H). Twenty-seven of 49 (55%) pwCF suffered from chronic rhinosinusitis (CRS) according to the European Position Paper on Rhinosinusitis and Nasal Polyps (EPOS) criteria including nasal blockage and discharge, postnasal drip, fascial pain, headache, and loss/reduction in sense of smell (47). Lung function testing during the study period resulted in a mean FEV₁ of 79.6 ± 30.9% (range, 21.5%–125.7%) during stable phases and 72.3 ± 30.2% (range, 22.0%–116.8%) during exacerbations. Twenty of 49 (40.8%) pwCF were intermittently or permanently colonized with P. aeruginosa, 16/49 (32.7%) were tested positive during trial, and 12/20 (60%) were adults. As a specific standard of the CF center, patients who tested positive for P. aeruginosa received pulmonary applied antibiotics for 1 year, following the last positive culture. Thirty-five of the 49 (71%) patients had a known S. aureus colonization; 28/35 (80%) of those were children. Ten of the 49 (20%) patients were co-infected in upper and lower airways with both pathogens. Patients colonized with P. aeruginosa received oral azithromycin and/or inhaled colistin, tobramycin, or aztreonam lysine as a standard therapy, according to the current standards of care (18). None of the patients received CFTR modulating therapy, as the study was completed before approval in Germany. Demographic data of the included patients are shown in Table 1 and in Supplementary Tables S1, S2 .

During the 13-month trial, we found a mean of 1.5 ± 0.6 exacerbations in the CF cohort and 1.4 ± 0.8 ARIs in the control cohort. There was no significant difference between both cohorts, but as expected, children revealed to have slightly higher infection rates than adults.

PCR analyses were performed in 213 of the 214 NLF samples collected over the 13-month study. During that period, out of the 14 viruses targeted by PCR, 5 were detected in the CF cohort (hRV, adenovirus, enterovirus, parainfluenza virus type 2, and hMPV), and 4 four were detected in the HC cohort (hRV, adenovirus, hMPV, and coronavirus). Among the n=31 (63.3%) pwCF who tested positive for any virus at least once during the whole observation period, n=23 (46.9%) tested positive for a single virus, n=6 (12.2%) for two different viruses, and n=2 (4.1%) for three different viruses. Over the whole study, 18 (36.7%) pwCF remained negative for all the targeted viral pathogens. Regarding the HC cohort, n=19 (50%) out of the 38 considered subjects tested positive for a single virus and n=9 (23.7%) for two different viruses. There were no subjects in the HC cohort with detection of more than two different viruses during the whole observation time frame, and n=10 (26%) individuals remained negative for all the viruses targeted in this study. Individuals infected with a viral pathogen at least once during the whole study were significantly younger compared to subjects who remained free of such pathogens (median, IQR: 14.3 [5.8, 27.3] years vs. 24.5 [15.7, 37.4] years; p=0.02).

The most frequently detected viral pathogen over the whole study was hRV in both groups, being detected at least once in n=24 (49.0%) of the 49 included pwCF and in n=26 (68.4%) of the 38 healthy individuals included in the study. In general, median age in HC subjects (median, IQR: 23.4 [4.4, 32.3] years) who became infected with hRV at least once during the whole study was greater than in pwCF (median, IQR: 11.3 [8.6, 25.0] years), but such a difference was not statistically significant.

At stable phases, hRV detection rates were equally distributed in both groups (18.4%). In contrast, during the first exacerbation, hRV detection rates were significantly higher in the HC group than in the CF group (65.8% vs. 22.4%; p<0.0001). Odds ratio calculations revealed that, at exacerbation, the probability for hRV to be detected in the healthy cohort was 6.6-fold higher than in the CF cohort [Wald confidence interval: (2.6, 17.2); p=0.005]. At stable phases and during exacerbation, the median age in the subgroup of individuals who tested positive for hRV was significantly lower than that in hRV-negative individuals (stable phase median, IQR: 8.9 [4.5, 21.1] years vs. 23.1 [9.7, 31.8] years, p=0.043; exacerbation median, IQR: 13.8 [4.7, 26.4] years vs. 23.1 [10.2, 35.3] years, p=0.018). At subsequent exacerbation time points, differences in hRV detection rates and in infection ages did not attain significance due to the small number of collected samples at those time points ( Supplementary Table S8 ).

Altogether, adenoviruses and enteroviruses were detected less frequently ( Supplementary Table S8 ). Furthermore, analyses of hRV detection with regard to azithromycin intake revealed that pwCF not receiving azithromycin (n=34) had a much higher probability for hRV detection compared to pwCF receiving azithromycin (n=15) (odds ratio, 5.1; p=0.033).

Regarding bacterial pathogens targeted by PCR, only one subject in the CF cohort tested positive for M. pneumonia during exacerbation, whereas during stable phases, no subject tested positive for those pathogens.

Given that not all participants were able to provide further NLF samples at the second and subsequent exacerbation time points, inflammatory markers analyses were restricted only to the first exacerbation time point, which is referred to only as exacerbation throughout the next sections in the paper.

CF patients revealed higher baseline levels of inflammatory cytokines during stable phases, and during exacerbation, cytokines did not increase to the level observed in healthy controls. This contributes to previous publications from our group (16, 49) and others (50, 51) showing the chronic inflammatory process typical for CF during stable phase. Nevertheless, cytokine levels during exacerbation were comparable in both groups, and stated differences did not reach statistical significance. In addition, children and adults showed similar inflammatory responses in both groups.

In this study, using multiplex PCR, viral pathogens were significantly more often detected during exacerbatioon/ARI in both cohorts. This is consistent with previous studies reporting viral infections to be responsible for up to 72% of exacerbations in pwCF (25). Furthermore, hRV resulted to be the most frequently detected virus in both CF and HC cohorts. This accords well with previous studies (24, 52–54) in which hRV has been reported as the most prevalent virus also in pwCF. In our study, however, we found that hRV detection rates in pwCF and HCs were similar during stable phases (18.4%), whereas during exacerbation, these rates were significantly higher in HC than in pwCF (65.8% and 22.4%, respectively). Discrepancies between our results and those reporting higher incidences of hRV infections in pwCF are likely due to the younger age of the cohorts considered in those studies. In our study, median age for hRV infection at exacerbation was 13.8 and 23.2 years (p=0.018) for the CF and HC cohorts, respectively.

At stable phases, hRV detection led to significantly higher levels of IL-6 and IL-8 in the whole cohort, (i.e. CF + HC), whereas in the hRV-positive subgroup of pwCF, only IL-6 was elevated. On the other hand, in the whole cohort, virus detection during exacerbation was associated with increased levels of IL-1β, IL-6, IL-8, NE, and MMP-9. Increases in IL-6, IL-8, NE, and MMP-9 were also observed in the subgroup of pwCF revealing viral colonization. This effect is likely due to the predominant effect of hRV colonization in our cohort, as its detection led to higher levels of IL-1β, IL-6, IL-8, NE, and MMP-9, whereas none of the other viruses different from hRV affected the levels of any assessed inflammatory parameter.

Host immune responses to hRV have been previously studied in vitro and in vivo (26, 55). The single-stranded RNA virus is known to infect the airway epithelium by binding to its target receptor, which varies according to the virus’ subtype (56). After infection, the airway epithelium responds by activating different signaling pathways, leading to the release of interleukin IL-8, regulated on activation, normal T cell expressed and secreted (RANTES, also known as CCL5), and granulocyte-macrophage colony-stimulating factor (GM-CSF), which in turn again recruits neutrophils (55). Other studies in pwCF reported associations between hRV infections and elevated IL-6, IL-8, and neutrophil levels (26, 55).

Regarding bacterial pathogen detection, in our study, only M. catarrhalis was found to be significantly associated with the elevation of inflammatory mediators in NLF. In a previous study, Moraxella species were found to be strongly associated with higher incidences of respiratory viruses in bronchoalveolar lavage fluid samples from pwCF (33). Nevertheless, in our study, detection rates for this bacterium were comparatively low (n=4), and only one of the hRV-positive patients was simultaneously positive for colonization with M. catarrhalis. Therefore, we cannot draw any conclusions regarding the previously reported association.

Medication with inhaled antibiotics (colistin, tobramycin, or aztreonam) and oral therapy with azithromycin significantly reduced all inflammatory cytokines and mediators except for TIMP-1 during exacerbation. This is consistent with recent studies indicating benefits of azithromycin with regard to decreased exacerbation rates (57) and lung function decline, visible also in reduced inflammation mediators (21, 36). It is remarkable that both drugs also show effects in the UAW and even in nasal lavage fluids, although they were orally and pulmonary administered. Azithromycin is known as a non-specific inhibitor of the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB) pathway, thus acting as an anti-microbial and anti-inflammatory agent (21, 58). Although overall mechanisms remain elusive, it is known that bacterial infections activate NF-kB pathway via toll-like receptors and airway epithelial cells, which, in turn, promotes transcription of pro-inflammatory substances like IL-6, IL-8, and interferon gamma (IFN-γ). Elevated IL-8 levels stimulate NE release in PMN. NE in turn is a major inhibitor of TIMP-1. In our study, TIMP-1 levels remained comparable in both groups, indicating reduced NE activity. As TIMP-1 plays an important role in inhibiting airway remodeling, therapy with azithromycin seems to be essential in preventing respiratory endothelial destruction in the UAW of CF patients. Furthermore, co-medication with inhalative antibiotics and azithromycin shows a synergistic effect, significantly reducing IL-6, IL-8, and NE levels also during stable phase. It therefore seems advisable to apply both drugs simultaneously.

Consistent with age-related colonization patterns of CF patients in different age groups in German and international registries (19), CF children recruited in this study were significantly more often colonized with S. aureus, whereas adults were more likely to have a P. aeruginosa colonization.

Colonization with P. aeruginosa was associated with lower baseline levels of inflammatory cytokines and further mediators like NE and MMP-9. In concordance with previous publications (21, 22), we attribute this to prescribed anti-inflammatory medication with azithromycin, as all patients with chronic P. aeruginosa colonization included in this study received azithromycin orally. As azithromycin directly inhibits the NF-κB pathway (59), which plays an important role in upper airway inflammation, reduced cytokine levels can be directly associated with this medication. We should highlight that this effect was more pronounced during the stable phase than during exacerbation.

Colonization of the UAW with S. aureus appears to contribute to the inflammatory process in CF to a higher extent than P. aeruginosa colonization (under antimicrobial medication), as patients colonized with S. aureus tended to show higher inflammatory mediators with significantly higher levels in IL-6, NE, and MMP-9 during the stable phase. This corresponds to previous findings by Janhsen et al. (22). S. aureus possesses different virulence factors like staphylococcal protein A that interact directly with tumor necrosis factor-alpha (TNF-α) receptor leading to a massive release of pro-inflammatory cytokines like IL-1β, IL-6, and IL-8 (60). Further staphylococcal proteins are enterotoxins A and B (SEA and SEB) activating T-cell proliferation and increase cytokine production (61). This pro-inflammatory effect seems to be more visible during the stable phase, as we observed comparable overall cytokine levels in both groups during exacerbation. Beside cytokine release, S. aureus apparently also promotes PMN degradation, as we observed elevated MMP-9 and NE levels during the stable phase and both mediators are PMN-secreted. Nowadays, it is assumed that PMNs play a major role in LAW immune system. UAW immune system is considered to be mostly IgA triggered (15, 34, 60, 62). However, elevated levels in MMP-9 and NE indicate a participation of PMNs also in UAW.

Although current guidelines do not advise permanent antibiotic medication in CF patients colonized with S. aureus (18), S. aureus colonization promotes UAW inflammation and contributes to disease progression.

Similar to single colonization with S. aureus, co-colonization with S. aureus and P. aeruginosa shows significantly higher levels of IL-6, NE, and MMP-9 during stable phases, and NE remains elevated during exacerbation. This indicates a dominant influence of S. aureus pathogens in co-colonized CF patients. According to Limoli et al. (63), lung function decline and increased exacerbation rates are associated with co-infections in CF patients. CF patients colonized with P. aeruginosa are regularly treated with oral azithromycin, a medication that S. aureus is known to be resistant to. Higher inflammatory responses in co-infected patients can thus be directly related to less effective medication. Hence, we should consider a more aggressive treatment including anti-staphylococcal antibiotics in co-infected CF patients with clinical lung deterioration.

In six samples, microbiology could not be performed due to an insufficient amount of material. Instead, for these samples, multiplex PCR analysis of viral and pneumonia-causing pathogens were conducted. Some subjects in the healthy control cohort were medical students or staff from the Jena University Hospital. In addition, bacterial colonization for the healthy cohort was not assessed, which may have limited the interpretation of results from comparisons with pwCF, as it is known that health workers may be at higher risk of colonization with MSSA and MRSA. However, rates of MRSA colonization are relatively low in our hospital, compared to other countries, so that we do not expect this to be a relevant confounder.