In this cross-sectional study, sinonasal tissues were collected from CRSwNP patients and control subjects from February 2014 to April 2017 by 4 well-trained experienced surgeons at the Samsung Medical Center, Seoul, Republic of Korea. The diagnosis of CRS was established based on the 2012 European position paper on rhinosinusitis and NP guidelines.1 Patients who underwent FESS once were defined as having primary NP (pNP). Seventy subjects who underwent primary FESS were selected based on the absence of recurrent polyps at subsequent outpatient visits with a follow-up time of over 24 months. Patients who were uncontrolled with medications, including antibiotics and corticosteroids, and then underwent FESS two or more times, were defined as having refractory NP (rNP). Usually, patients were prescribed oral corticosteroids (prednisolone, 10–30 mg per day) for 1–2 weeks or oral antibiotics more than two weeks based on a bacterial culture result. Some of the asthmatic patients should be pre-medicated with several days' systemic corticosteroids to prevent anesthesia or surgery-induced pulmonary complications according to the local Institutional Review Board (IRB) guidelines and hospital policy. Sinonasal NP tissues from rNP patients who had bilateral diffuse polyposis were collected in the middle meatus at the time of revision surgery. Patients who underwent skull base surgery or endoscopic dacryocystorhinostomy with no signs or symptoms of nasal inflammation were included as control subjects. Patients were excluded if they were at less than 18 years of age or had unilateral disease, primary ciliary dyskinesia, cystic fibrosis, or eosinophilic granulomatosis with polyangiitis. Atopic status was evaluated by screening for serum specific immunoglobulin (Ig) E antibodies to common aeroallergens using ImmunoCAP^® assays (Phadia AB, Uppsala, Sweden) according to the manufacture's recommendations. A diagnosis of asthma was made by an allergist based on spirometry (change in FEV1 of 12% or more following administration of a short-acting β-agonist) or methacholine challenge test. Endoscopic examination was used to obtain the Lund-Kennedy score and computed tomography (CT) images were staged with the Lund-Mackay score.1415 Bacterial cultures were taken when mucopurulence was observed during endoscopy. Demographic details of subject's characteristics are shown in Table. CRSwNP was classified as either eosinophilic NP (ENP) or non-ENP (NENP), where tissue eosinophilia was defined as an eosinophilic count exceeding 10% of the total inflammatory cells per high power field in NP tissue.16 The IRB of Samsung Medical Center approved this study (IRB No. SMC 2017-02-129) and informed consent was obtained. Biospecimens for this study were provided by the Samsung Medical Center Biobank.

NP tissues from CRSwNP and UP tissues from control subjects were obtained. Tissues were homogenized and supernatants were stored at −80ºC until analyzed as previously described.171819 Protein concentrations for tissue extracts were determined using a Pierce 660-nm Protein Assay Kit (Thermo Scientific Inc., Rochester, NY, USA). Samples were thawed at room temperature and vortexed to ensure complete mixing. Multiplex cytokine analysis kits (B-cell activating factor [BAFF], C-C motif chemokine ligand [CCL]-11, CCL-24, interleukin (IL)-5, IL-8, IL-13, IL-17A, IL-23, interferon [IFN]-γ, myeloperoxidase [MPO], matrix metalloproteinase [MMP]-1, MMP-2, MMP-3, MMP-7, MMP-9, tissue inhibitor of metalloproteinase [TIMP]-1 and transforming growth factor [TGF]-β1) were obtained from R&D systems (cat. No. LMSAHM) and data were collected using a Luminex 100 (Luminex, Austin, TX, USA). Data analysis was performed using MasterPlex QT V.2.0 (MiraiBio, Alameda, CA, USA). All assays were run in duplicate according to the manufacturers' protocol. The sensitivity of each cytokine was as follows: BAFF (1.01 pg/mL), CCL-11 (14.6 pg/mL), CCL-24 (1.34 pg/mL), IL-5 (0.5 pg/mL), IL-8 (1.8 pg/mL), IL-13 (32.4 pg/mL), IL-17A (1.8 pg/mL), IL-23 (11.4 pg/mL), IFN-γ (0.4 pg/mL), MPO (20.4 pg/mL), MMP-1 (2.7 pg/mL), MMP-2 (108 pg/mL), MMP-3 (5.3 pg/mL), MMP-7 (23.2 pg/mL), MMP-9 (13.6 pg/mL), TIMP-1 (3.42 pg/mL) and TGF-β1 (2.1-24.6 pg/mL). Mucosal tissue eosinophil cationic protein (ECP) were measured by ImmunoCAP® and the sensitivities for ECP were 2 μg/L. Anti-dsDNA quantitative IgG and IgA enzyme immunoassays (EIAs; Alpco Diagnostics, Salem, NH, USA) were also assessed. The detection range for all EIAs was 30 to 150,000 IU/mL. Protein levels in tissue homogenates were normalized to the concentration of total protein (mg/mL).

IHC was performed with the Polink-2 plus polymerized horseradish peroxidase (HRP) broad DAB Detection System (Golden Bridge International Labs, Bothell, WA, USA) as previously described.20 Briefly, after deparaffinization sections were incubated in 3% hydrogen peroxide to inhibit endogenous peroxidase activity. Heat-induced epitope retrieval was performed by microwaving samples in 10 mmol/L citrate buffer (pH 6.0). Sections were then incubated for 60 min at room temperature with primary antibody consisting of mouse anti-human neutrophil elastase (HNE; 1:500; R&D Systems, Minneapolis, MN, USA). The sections were then incubated in a broad antibody enhancer, followed by polymer-HRP and stained using the DAB Detection System reagent. Finally, slides were counterstained with hematoxylin.

Data were tested for normality of distribution using the Shapiro-Wilk test and are reported as median with interquartile range due to a nonparametric distribution. Data were analyzed by the Kruskal-Wallis test with the Dunn multiple comparison test. Tests for statistical differences for binary classification were analyzed using Fisher's exact test. Correlations were tested using Spearman's rank correlation coefficients. A P value of less than 0.05 was considered statistically significant. Principal component analysis (PCA) was used to describe patterns of inflammatory mediators. Analyses were performed using Stata software v14.0 (StataCorp, College Station, TX, USA) and Graphpad prism software 6.0 (GraphPad software Inc, La Jolla, CA, USA).

Refractory CRSwNP patients underwent revision surgery despite primary FESS and postoperative medical treatments have more extensive disease, worse postoperative courses and were younger compared to patients with primary CRSwNP (Table). In addition, preoperative endoscopic and CT scores were higher in rNP than in pNP. Compared to pNP subjects, rNP subjects showed higher infection rate (mucopurulence and culture-positive rate) and more frequent use of antibiotics and suffered from symptomatic bacterial infection.

To characterize the inflammatory profiles of refractory CRSwNP patients, multiplex bead-based immunoassay for inflammatory cytokines was performed in 70 pNP, 86 rNP and 23 control subjects (Fig. 1). Th2 inflammatory markers, such as IL-5, ECP, CCL-11, CCL-24 and IL-13, were up-regulated in pNP and rNP, but not significantly different between pNP and rNP subjects. Interestingly, levels of Th1 cytokine (IFN-γ), BAFF and Th17-associated mediators, such as MPO, IL-8, IL-17A and IL-23, were increased in rNP compared to pNP. With respect to remodeling markers, both pNP and rNP exhibited lower levels of TGF-β1 compared to controls and levels of TGF-β1 were lower in rNP than in pNP. MMP2/TIMP1 and MMP9/TIMP1 ratios were all up-regulated in rNP compared to controls and pNP.

We next asked whether levels of inflammatory mediators could make a discrimination between rNP and pNP in each subtype. To do so, a multivariate analysis of multiplex data was performed by exploratory factor analysis using PCA of the correlation matrices of major mediator levels measured (ECP, BAFF, CCL-24, IL-5, IL-8, IL-13, IL-17A and IFN-γ). PCA retained 2 components explaining 47.1% of all variance in the data (Fig. 2). The first component (component 1) accounted for 26.3% of the variance and consisted of IFN-γ, BAFF and IL-17A in decreasing order. The second component (component 2) accounted for 20.8% of the variance and was composed of CCL-24, ECP and IL-5. Component 1 was predominantly represented by BAFF and Th17/Th1related markers, whereas component 2 was consistent with a Th2 profile (Supplementary Tables S1 and S2). When factor scores were analyzed using the Mann-Whitney test, components 1 and 2 were significantly different among groups (data not shown). We also found that component 1 was able to discriminate between pNP and rNP in both eosinophilic and non-eosinophilic subtypes. These findings demonstrated that BAFF- and Th17/Th1-related markers are the main biomarkers for characterizing rNP regardless of tissue eosinophilia. Interestingly, neutrophilic infiltration in refractory ENP (rENP) and refractory NENP was also enhanced based on IHC using anti-HNE compared to pNP (Fig.3). Eosinophilic infiltration was increased in pNP and rNP compared to control tissues, whereas no difference was observed between pNP and rNP (Fig. 3).

To investigate the differences in the inflammatory profiles of NP in asthmatic rNP versus non-asthmatic rNP, we used a multiplex bead-based immunoassay for inflammatory cytokines in 20 asthmatic rNP and 66 non-asthmatic rNP (Fig.4). MPO, IFN-γ, IL-17A and BAFF were increased, therefore, appear to be common signature mediators in both non-asthmatic and asthmatic rNP cases. Moreover, ECP and CCL-24 were up-regulated in asthmatic rNP compared to pNP and controls. Eosinophilic inflammation may affect the refractoriness of the disease in asthmatic rNP patients.

High levels of BAFF are associated with autoimmune diseases; additionally, NP tissues from patients with revision surgery showed higher levels anti-dsDNA IgG antibody compared to primary surgery tissues.21 Therefore, we measured levels of anti-dsDNA IgG and IgA antibodies levels in pNP and rNP (Fig. 5). The rNP demonstrated a higher level of anti-dsDNA IgG and IgA compared to controls. In subgroup analysis, asthmatic rENP showed higher levels of anti-dsDNA IgG and IgA antibodies compared to the other groups (Fig. 5).