Abstract
Background:
Multiple hypotheses are evolving which suggest several, potentially overlapping etiologies for olfactory dysfunction (OD) in chronic rhinosinusitis (CRS). Understanding inflammatory cytokine profiles of the olfactory cleft (OC) and their association with olfactory function is foundational for future clinical care and research.
Methods:
This cross-sectional, case control study evaluates associations amongst OC mucus inflammatory proteins, psychophysical olfactory testing, and computed tomography (CT) analysis of the OC and sinuses. Normative reference intervals were determined for each protein and odds ratios (ORs) used to compare proportions of altered expression between CRSsNP and CRSwNP.
Results:
Case subjects with CRS (n=151) and controls (n=74) were evaluated. A majority of OC proteins tested were found within detectable ranges for cases and controls. The CRS cohort had significantly higher concentrations for 23 of 26 proteins. CRS cases with abnormal levels of CCL2, CCL3, IL5, IL10 and IL13 associated with greater olfactory deficits. The prevalence of elevated IL5 and IL13 in anosmic patients was 64.6% and 62.5% respectively (p<0.004). CRS cases with the highest odds of elevated expression in CRSwNP were IL5 (OR=10.83) and IL13 (OR=8.36). However, both IL5 and IL13 were still elevated in approximately 14% of CRSsNP patients. The highest magnitude of correlation between the total percent of OC opacification was found to be with IL5 (r=0.543; p<0.001) while other moderate correlations were noted with IgE, IL10 and IL13.
Conclusions:
This study confirmed that OC inflammatory proteins vary both by disease phenotype and in their association with OD. Type-2 inflammatory mediators are increased in CRS, especially within the CRSwNP group. However, a substantial proportion of CRSsNP also express type-2 inflammatory mediators. Further research is necessary to understand the complex roles OC mucous inflammatory proteins might play in defining endotype and in impacting CRS-related OD.
MeSH Key Words: Outcome assessment (health care), sinusitis, chronic disease, smell, proteins, inflammation, olfaction
INTRODUCTION
Chronic rhinosinusitis (CRS) is one of the most common causes of olfactory dysfunction (OD) which is present in 60–80% of patients with CRS and is considered a cardinal diagnostic feature. Historically, CRS-OD has been primarily associated with polyposis and the prevailing hypotheses have revolved around olfactory cleft (OC) obstruction to airflow. More recently, multiple hypotheses are evolving to suggest that there may be several, potentially overlapping etiologies for OD in CRS. These include direct inflammation of the olfactory neuroepithelium, metaplasia of the neuroepithelium, and/or distorted OC mucus physiology.1
The ability to study the pathophysiology of OD has greatly improved in recent years allowing study of a neuroepithelium that is challenging to biopsy effectively and safely.2–5 Recent techniques to collect OC mucus proteins in a nondestructive way have been successfully implemented using filter paper placement directly into the OC under endoscopic visualization.1,6,7,8 Prior studies limited to CRS patients reveal that several OC and middle meatus proteins correlate with olfactory function and OC opacification measured radiologically,1,8 however these studies lack control subjects and normative ranges of these proteins are unknown. Lack of normative values has impaired our ability to categorize patients within a given phenotype to specific endotypes. This type of personalized information is critical as we enter the era of biologics that target specific inflammatory cytokines. Thus, the approach for the current study is to expand data collection to a larger and more diverse patient population from multiple centers, including control subjects, in order to define normative ranges for each cytokine. This enabled us to pursue the goal of identifying the proportion of CRS patients with abnormal levels of each protein and the association with individual patient olfactory function.
Our primary hypothesis is that disease-specific differences between OC proteins will exist between control subjects and subjects with CRS and that this will be primarily driven by increased Type-2 cytokines in both absolute levels and proportion of patients affected. Our secondary hypothesis is that certain OC protein concentrations will correlate with olfactory function measured by psychophysical testing. Better understanding of the mechanisms underlying CRS-OD and personalized therapeutic options will begin to enable a paradigm shift in the clinical management of OD in CRS.
MATERIALS and METHODS
Study Patient Population with CRS
Case subject enrollment originated from an observational, prospective research investigation of human subjects funded by the National Institute on Deafness and Other Communication Disorders (Bethesda, MD.; Federal grant 3R01 DC005805). Case study participants were recruited from patient populations presenting to academic, rhinology centers located in the United States within: Oregon Health & Science University (Portland, OR), the Medical University of South Carolina (MUSC, Charleston, SC), the University of Utah (Salt Lake City, UT), the University of Colorado (Aurora, CO), and the University of Virginia (Charlottesville, VA). Symptomatic, adult study participants (≥ 18 years of age) received a confirmed diagnosis of symptomatic and medically recalcitrant CRS, with nasal polyposis (CRSwNP) or without nasal polyposis (CRSsNP), from a fellowship-trained rhinologist following criteria established by current practice guidelines.9 All study participants provided written, informed consent after enrollment meetings to ensure voluntary study participation. Of note, patients were permitted to be on baseline maintenance medical therapy for CRS with the exception of oral corticosteroids at the time of sampling. This reflects the clinical situation where we expect endotyping would be employed in the future for clinical decision-making.
Control Population without CRS
Control participants were enrolled from a community-based population sample of healthy subjects without signs, symptoms or history of sinonasal disease. Control participants were prospectively interviewed and enrolled on a voluntary basis. Adult, study volunteers were recruited locally using advertisements, word-of-mouth, and self-referral. The Institutional Review Board affiliated with each enrollment location approved study protocols while providing annual review for patient and data safety. All study participants provided a medical and social history during baseline enrollment meetings.
Exclusion Criteria
Participants were excluded if medical history indicated a history of ESS within the past 6 months or comorbid conditions associated with increased prevalence of OD at the time of enrollment including: Sarcoidosis, granulomatosis with polyangitis, cystic fibrosis/ciliary dysfunction, Alzheimer’s disease, Parkinson’s disease, other non-specified neurocognitive disorders, major head trauma / traumatic brain injury, and any immunosuppressive medication.
Radiographic and Endoscopic Evaluations
Measures of disease severity were collected for diagnostic purposes and sourced as research data when provided by the standard of care for case subjects. Patients with CRS underwent sinonasal computed tomography (CT). Image findings were quantified using the bilateral Lund-Mackay (LM) staging system (score range: 0–24).10
Further volumetric analysis of CT images was completed using OsiriX MD imaging software (version 9.0.1; Pixmeo, Bernex, Switzerland). Digital Imaging and Communication in Medicine images were uploaded, and 2D segmentation of the olfactory cleft was performed manually in OsiriX (Pixmeo) using the bone window in the coronal plane. Anterior, middle, and posterior 2D cross-sectional areas were demarcated in the coronal plane using the axial and sagittal planes for reference of anatomic landmarks. The quantitative opacification was determined using pixel-based software analyses. The total average percentage (%) of opacification of the OC was then determined by averaging across anterior, middle, and posterior regions. Each scan was reviewed by two trained research personnel who were blinded to olfactory assessment scores.11 Higher total percentages reflect worse overall obstruction in the OC region.
Olfactory Assessments
Case and control subjects completed evaluations of bilateral psychophysical olfactory function using Sniffin’ Sticks pens (Burghart Messtechnik, Wedel, Germany) which evaluate three separate domains of olfactory function including: odorant threshold (score range: 1–16), odorant discrimination (score range: 0–16), and odorant identification (score range: 0–16).12,13 The TDI total is commonly interpreted by a corresponding diagnostic category of either anosmia (score range: 1–15), hyposmia (score range: 16–30), and normosmia (score range: 31–48).14
Olfactory Cleft Mucus Collection and Laboratory Analyses
All study subjects consented to OC mucus collection as a condition of study participation. Nasal endoscopy was performed on each study subject using rigid endoscopy (Karl Storz, Tuttlingen, Germany) after application of topical lidocaine and phenylephrine to the nasal cavity via atomizer. Under direct visualization, a sterile filter paper (Leukosorb; Pall Scientific, Port Washington, NY) strip was placed directly into the OC bilaterally.15 The Leukosorb paper was then removed and placed into cuvettes that were centrifuged at 4°C for 10 minutes at 10,000rpm. The mucus was then combined from each side, transferred by pipette to a cryotube, and stored at −80°C until specimens were transferred to MUSC for performance of assays.
An array of growth factors, cytokines, and chemokines was chosen for analysis based on previous evidence, suggesting a role in CRS, olfactory dysfunction, or inflammation/remodeling. The manufacturer’s instructions were followed to execute the CBA assay and as we have previously described.1,6 In addition to the manufacturer’s instructions, a detailed Legendplex assay protocol can be found at https://www.jove.com/t/56440/multiplex-cytokine-profiling-stimulated-mouse-splenocytes-using. Nasal mucus samples were immediately run following the assay conclusion on a Guava 8HT flow cytometer, collecting 5,000 events/bead in the plex. Data was analyzed using the Legendplex Data Analysis software. All results are presented as concentration of cytokines in picograms per milliliter of nasal mucus, thus eliminating the possibility of volume influencing the outcome. Assay dilute served as a negative control and the recombinant proteins used in the generation of the standard curve served as a positive control. Detection limits were specific to each bead and were reported following the recommendation of the manufacture and Legendplex Data Analysis software. Mucus volumes were not specifically measured in this study. However, in a previous unpublished study, we found no difference in nasal mucus volumes between control subjects and individuals with CRSsNP or CRSwNP.
Total immunoglobulin E (IgE) was quantified using enzyme-linked immunoassay following the kit instructions (GenWay Biotech, Inc, San Diego, CA).
Whenever values were above or below the detectable limit, the highest or lowest recorded value was imputed. Assays were performed on all samples with an adequate volume of mucus collected. Variation in sample sizes due to specimen volume differences for each OC mucus protein are indicated in each table.
Data Management and Statistical Analyses
Data security was ensured through the assignment of unique study identification numbers for study participants and removal of all protected health information prior to data or specimen transfer. A closed-environment database was used for data collection (Access; Microsoft Corporation; Redmond, WA) and statistical analyses were completed using SPSS software (ver. 26.0; IBM Corporation, Armonk, NY). Between-group comparisons of patient characteristics, comorbidity at enrollment, or olfactory category/diagnoses were completed using either independent samples t-testing or chi-square (χ2) testing for proportional differences. Between-group comparisons of cytokine expression data were conducted using Mann-Whitney U testing for rank sums due to the non-Gaussian distribution of independent cytokine expression data.
Reference intervals (RIs) were also derived from cytokine expression data using findings from the reference control population. These RI values were determined using a non-parametric approach by excluding the upper 2.5% and lower 2.5% of control subjects to define a 95% RI.16–19 Abnormal cytokine expression was identified in case subjects with CRS if expression levels fell outside of that RI range. Odds ratios (ORs) were used to compare the prevalence of normal and abnormal expression between CRSsNP and CRSwNP. Additional bivariate association between any two scaled measures utilized two-tailed Speaman’s rank correlation coefficients (R). All testing statistics / effect estimates, 95% confidence intervals (CIs), and both adjusted and unadjusted type-I error probabilities (p-values) are reported. Family-wise error rates were adjusted using the Benjamini-Hochberg procedure (q-values), where appropriate, which employed a conservative 5.0% false discovery ratio (FDR) to control false positives identified within between-group comparisons of mucus olfactory protein concentrations.
RESULTS
Final Study Populations
A total of 225 study participants met final study inclusion criteria and provided consent between December, 2016 and May, 2019 consisting of 151 (67%) case subjects with CRS and 74 (33%) control subjects without CRS. Study participant demographic factors, comorbidity, clinical measures of disease severity and olfactory assessment scores are described in Table 1 including characteristics of case subjects with either CRSsNP (n=64) or CRSwNP (n=87).
Table 1:
Overall characteristics of final study populations at enrollment (n=225)
| Demographics: | All Case Subjects with CRS (N=151) |
CRSsNP (N=64) |
CRSwNP (N=87) |
Control Subjects (N=74) |
Test statistic | p-value |
|---|---|---|---|---|---|---|
| Age in years Mean [±SD] | 48.4 ± 15.9 | 48.1 ± 16.3 | 48.6 ± 15.7 | 50.4 ± 18.0 | t=0.85 | 0.394 |
| Males N (%) | 70 (46%) | 30 (47%) | 40 (46%) | 27 (37%) | χ2=1.97 | 0.160 |
| Females | 81 (54%) | 34 (53%) | 47 (54%) | 47 (63% | ||
| White/Caucasian | 130 (86%) | 61 (95%) | 69 (79%) | 55 (74%) | χ2=4.72 | 0.194 |
| African American | 19 (13%) | 2 (3%) | 17 (20%) | 17 (23%) | ||
| Asian | 1 (1%) | 1 (2%) | 1 (1%) | 1 (1%) | ||
| Hispanic/Latino ethnicity | 8 (5%) | 5 (8%) | 3 (3%) | 3 (4%) | χ2=0.17 | 0.684 |
| Comorbidity: | ||||||
| Nasal polyposis | 87 (58%) | 0 (0%) | 87 (100%) | 0 (0%) | χ2=69.52 | <0.001 |
| Previous sinus surgery / ESS | 76 (50%) | 23 (36%) | 53 (61%) | 0 (0%) | χ2=56.24 | <0.001 |
| Asthma | 62 (41%) | 22 (34%) | 40 (46%) | 1 (1%) | χ2=38.84 | <0.001 |
| Diabetes mellitus (Type I/II) | 15 (10%) | 9 (11%) | 6 (7%) | 5 (7%) | χ2=0.62 | 0.431 |
| Smoking / tobacco use (current) | 5 (3%) | 2 (3%) | 3 (3%) | 4 (5%) | χ2=0.57 | 0.451 |
| Positive allergy test (mRast/skin prick) | 81 (54%) | 31 (48%) | 50 (58%) | 17 (23%) | χ2=19.00 | <0.001 |
| Allergic fungal rhinosinusitis | 10 (7%) | 0 (0%) | 10 (12%) | 0 (0.0%) | χ2=5.13 | 0.024 |
| AERD | 21 (14%) | 6 (9%) | 15 (17%) | 0 (0.0%) | χ2=11.35 | 0.001 |
| Medication usage at sampling: | ||||||
| Topical corticosteroids (rinses/sprays) | 121 (81%) | 53 (83%) | 68 (80%) | 0 (0.0%) | χ2=131.38 | <0.001 |
| Antihistamines | 72 (48%) | 31 (48%) | 41 (48%) | 0 (0.0%) | χ2=52.81 | <0.001 |
| Saline irrigations | 105 (71%) | 48 (75%) | 57 (67%) | 0 (0.0%) | χ2=98.55 | <0.001 |
| Anti-leukotrienes | 34 (23%) | 12 (19%) | 22 (26%) | 0 (0.0%) | χ2=19.92 | <0.001 |
| Disease severity measures: | ||||||
| Lund-Mackay CT score | 13.4 ± 5.8 | 9.4 ± 5.0 | 16.5 ± 4.3 | ---- | ---- | ---- |
| Anterior OC opacification (%) | 57.0 ± 31.6 | 38.5 ± 23.8 | 70.9 ± 30.0 | ---- | ---- | ---- |
| Middle OC opacification (%) | 63.7 ± 29.0 | 47.7 ± 25.0 | 75.8 ± 26.0 | ---- | ---- | ---- |
| Posterior OC opacification (%) | 68.8 ± 28.8 | 53.4 ± 27.0 | 80.3 ± 24.4 | ---- | ---- | ---- |
| Sniffin’ Sticks total olfaction score | 22.1 ± 9.3 | 27.8 ± 6.8 | 17.9 ± 8.7 | 30.5 ± 5.7 | t=8.32 | <0.001 |
| Threshold score | 3.7 ± 2.9 | 4.8 ± 2.7 | 2.9 ± 2.7 | 6.7 ± 2.4 | t=8.27 | <0.001 |
| Discrimination score | 9.1 ± 3.4 | 11.2 ± 2.9 | 7.6 ± 2.9 | 11.6 ± 2.3 | t=6.27 | <0.001 |
| Identification score | 9.3 ± 4.3 | 11.8 ± 2.9 | 7.4 ± 4.2 | 12.2 ± 2.2 | t=6.67 | <0.001 |
LEGEND: CRS, chronic rhinosinusitis; CRSsNP, CRS without nasal polyposis; CRSwNP, CRS with nasal polyposis; SD, standard deviation; ESS, endoscopic sinus surgery; mRAST, modified radioallergosorbent testing; AERD, aspirin exacerbated respiratory disease; CT, computed tomography; OC, olfactory cleft; N, sample size; t, independent sampling t-testing; χ2, chi-square testing; Statistical comparisons were between all case subjects with CRS (n=151) and control subjects (n=74).
Overall Descriptive Findings for Olfactory Cleft Mucus Proteins
Descriptive findings for OC mucus proteins for both case and control subjects are provided in Table 2. Wide data variability reflected by high standard deviation values were associated with mean protein concentrations for all case and control subjects. The majority of OC mucus proteins were found within detectable ranges for case subjects with CRS with 15 of 26 proteins having ≥90% detectable samples. Comparably, the majority of specimens for control subjects were found within detectable ranges for 16 of 26 proteins with ≥90% with detectable samples. Notable exceptions in control specimens were CCL5 and IL2 proteins with ≥90% of specimens below the detectable limit.
Table 2:
Comparison of OC mucus protein concentrations between case subjects with CRS and controls and determination of normative reference interval
| Protein concentration All Case Subjects with CRS (pg/mL, n=151) |
Protein concentration Control Subjects (pg/mL, n=74) |
Mann-Whitney U test statistic | Unadjusted p-values |
q-values | ||||
|---|---|---|---|---|---|---|---|---|
| Mean ± SD | Median | Mean ± SD | Median | 95% reference interval | ||||
| CCL2 | 2456.5 ± 2997.9 | 1612.7 | 865.6 ± 850.9 | 612.7 | 211.4 – 3586.6 | 7.73 | <0.0001 | <0.0001 |
| CCL3 | 246.1 ± 838.0 | 53.5 | 11.5 ± 17.1 | 5.9 | 2.2 – 49.4 | 8.62 | <0.0001 | <0.0001 |
| CCL5 | 369.9 ± 2033.6 | 53.4 | 14.5 ± 38.2 | 6.2 | 6.1 – 109.7 | 10.72 | <0.0001 | <0.0001 |
| CCL11 | 609.4 ± 549.0 | 453.3 | 454.4 ± 432.3 | 361.1 | 50.1 – 1473.0 | 2.33 | 0.020 | 0.023 |
| CCL20 | 2894.5 ± 6002.6 | 692.3 | 309.1 ± 567.6 | 182.4 | 33.2 – 1189.4 | 5.99 | <0.0001 | <0.0001 |
| CXCL1 | 8660.8 ± 4475.1 | 7827.4 | 3940.7 ± 1019.0 | 4093.7 | 1901.4 – 5625.6 | 9.44 | <0.0001 | <0.0001 |
| CXCL5 | 4474.7 ± 4974.0 | 2330.9 | 2157.5 ± 2348.9 | 1545.7 | 114.1 – 9067.2 | 3.36 | 0.001 | 0.001 |
| CXCL9 | 3451.3 ± 3850.8 | 1997.3 | 1498.8 ± 2026.4 | 989.0 | 39.0 – 7183.1 | 4.45 | <0.0001 | <0.0001 |
| CXCL11 | 546.5 ± 2550.0 | 101.8 | 129.0 ± 327.1 | 28.4 | 3.7 – 664.5 | 5.41 | <0.0001 | <0.0001 |
| EGF | 794.7 ± 791.5 | 521.6 | 806.3 ± 567.2 | 626.5 | 219.9 – 2410.1 | −1.56 | 0.119 | 0.124 |
| bFGF | 2529.0 ± 4144.8 | 973.7 | 303.0 ± 388.8 | 190.5 | 43.7 – 1167.0 | 7.90 | <0.0001 | <0.0001 |
| IgE | 899.5 ± 2171.7 | 156.9 | 206.8 ± 295.4 | 74.1 | 11.6 – 911.6 | 2.50 | 0.012 | 0.014 |
| IL2 | 26.8 ± 60.9 | 10.0 | 8.6 ± 22.4 | 2.5 | 2.5 – 76.6 | 7.77 | <0.0001 | <0.0001 |
| IL4 | 10.0 ± 14.8 | 5.8 | 7.7 ± 15.4 | 3.4 | 1.9 – 49.4 | 4.62 | <0.0001 | <0.0001 |
| IL5 | 958.0 ± 2482.0 | 97.8 | 27.9 ± 87.0 | 6.3 | 2.3 – 333.7 | 7.22 | <0.0001 | <0.0001 |
| IL6 | 783.8 ± 1632.4 | 278.0 | 135.9 ± 408.1 | 35.5 | 4.9 – 704.4 | 8.39 | <0.0001 | <0.0001 |
| IL8 | 12637.2 ± 5833.4 | 11280.2 | 15628.6 ± 8682.0 | 15626.6 | 2400.8 – 29910.3 | −2.24 | 0.025 | 0.027 |
| IL9 | 39.3 ± 89.1 | 10.5 | 10.8 ± 28.7 | 4.1 | 1.9 – 47.6 | 5.48 | <0.0001 | <0.0001 |
| IL10 | 26.8 ± 180.4 | 4.3 | 1.5 ± 1.8 | 1.0 | 0.9 – 3.5 | 10.34 | <0.0001 | <0.0001 |
| IL13 | 201.6 ± 427.3 | 54.8 | 28.4 ± 49.9 | 15.4 | 3.1 – 78.1 | 5.46 | <0.0001 | <0.0001 |
| IL17A | 31.7 ± 162.3 | 12.4 | 3.1 ± 8.9 | 1.2 | 1.0 – 17.8 | 8.81 | <0.0001 | <0.0001 |
| IL17F | 21.1 ± 56.3 | 5.8 | 10.5 ± 28.4 | 2.7 | 1.2 – 77.9 | 6.89 | <0.0001 | <0.0001 |
| IL23 | 12.2 ± 20.2 | 8.6 | 4.8 ± 2.5 | 4.0 | 2.4 – 11.1 | 6.04 | <0.0001 | <0.0001 |
| IL33 | 5030.2 ± 16616.2 | 1292.1 | 1108.8 ± 1256.2 | 674.8 | 48.9 – 3884.5 | 3.85 | 0.00012 | 0.0002 |
| TNFα | 64.7 ± 326.2 | 9.1 | 2.8 ± 2.4 | 1.8 | 1.2 – 8.6 | 9.26 | <0.0001 | <0.0001 |
| VEGF-A | 4397.4 ± 1651.0 | 4097.7 | 4234.0 ± 1167.8 | 4208.8 | 1931.2 – 6381.4 | −0.11 | 0.913 | 0.913 |
LEGEND: All reported protein concentrations are expressed in picograms per milliliter (pg./mL). OC, olfactory cleft; CRS, chronic rhinosinusitis; N, sample size; SD, standard deviation; MWU, Mann-Whitney U test statistic; CCL, chemokine (C-C motif) ligand; CXCL, C-X-X motif chemokine; EGF, epidermal growth factor; bFGF, basic fibroblast growth factor; IgE, immunoglobulin E; IL, interleukin; TNFα, tumor necrosis factor alpha; VEGF-A, vascular endothelial growth factor-A. Q-values, Benjamini-Hochberg p-value adjustment using a 5% false discovery rate.
Between-Group Comparisons of OC Mucus Protein Concentrations
OC mucus protein concentrations were compared between case and control subgroups using nonparametric ranking statistics. The overall CRS cohort was found to have significantly higher concentrations for 23 of 26 of the proteins targeted in this study (p<0.050; Table 2). EGF and VEGF-A were not significantly different between groups while concentrations of IL8 were found to be significantly lower (p<0.030) in all case subjects with CRS. We then examined differences based upon phenotype as determined by presence or absence of NP. Patients with CRSsNP had significantly higher levels of all proteins compared to control subjects except for EGF, IgE, IL8 and IL13 (data in Tables 2 and 3, statistical analysis not shown). After adjustments for the FDR, patients with CRSwNP had significantly higher concentrations of CCL2, CCL3, CXCL11, IgE, IL5, IL9, IL13 than CRSsNP patients, but significantly lower IL8 and VEGF-A (p<0.050, Table 3). Concentrations of both IL10 and IL17A were significantly different between CRSsNP and CRSwNP before multiple comparisons correction (p<0.050). As expected, numerous type-2 cytokines were elevated in CRSwNP, although interestingly, IL4 was not different between groups.
Table 3:
Comparison of OC mucus protein concentrations between CRSsNP and CRSwNP case subjects
| Protein concentration CRSsNP (pg/mL, n=64) |
Protein concentration CRSwNP (pg/mL, n=87) |
Mann-Whitney U test statistic | Unadjusted p-values |
q-values | |||
|---|---|---|---|---|---|---|---|
| Mean ± SD | Median | Mean ± SD | Median | ||||
| CCL2 | 1422.3 ± 1015.8 | 1196.7 | 3217.4 ± 3679.4 | 2088.4 | 4.62 | <0.0001 | <0.0001 |
| CCL3 | 65.7 ± 101.4 | 31.2 | 378.7 ± 1084.1 | 94.7 | 4.30 | <0.0001 | <0.0001 |
| CCL5 | 231.0 ± 686.0 | 57.1 | 472.1 ± 2616.0 | 51.8 | −0.29 | 0.772 | 0.803 |
| CCL11 | 639.2 ± 528.9 | 450.6 | 587.4 ± 565.2 | 467.1 | −0.87 | 0.385 | 0.556 |
| CCL20 | 2216.2 ± 4626.3 | 744.4 | 3393.5 ± 6824.3 | 571.7 | −0.38 | 0.704 | 0.762 |
| CXCL1 | 8878.4 ± 4120.1 | 8404.6 | 8500.7 ± 4736.3 | 7360.7 | −1.08 | 0.279 | 0.483 |
| CXCL5 | 4365.3 ± 5193.7 | 2051.9 | 4555.2 ± 4834.8 | 2449.0 | 0.73 | 0.464 | 0.635 |
| CXCL9 | 3802.1 ± 4107.4 | 2220.9 | 3193.2 ± 3653.3 | 1858.4 | −1.05 | 0.293 | 0.476 |
| CXCL11 | 430.4 ± 679.5 | 176.3 | 631.9 ± 3314.5 | 72.6 | −3.32 | 0.001 | 0.0037 |
| EGF | 943.4 ± 996.2 | 532.8 | 685.3 ± 580.5 | 505.3 | −1.12 | 0.262 | 0.486 |
| bFGF | 2129.0 ± 3649.8 | 970.2 | 2823.3 ± 4472.0 | 1009.9 | 0.23 | 0.817 | 0.817 |
| IgE | 199.1 ± 336.5 | 104.4 | 1424.9 ± 2749.8 | 283.8 | 4.33 | <0.0001 | <0.0001 |
| IL2 | 37.6 ± 84.9 | 9.5 | 19.7 ± 36.3 | 10.0 | 0.57 | 0.570 | 0.644 |
| IL4 | 9.8 ± 15.0 | 5.8 | 10.2 ± 14.7 | 5.8 | 0.67 | 0.503 | 0.654 |
| IL5 | 164.4 ± 417.3 | 12.5 | 1487.0 ± 3081.4 | 500.4 | 6.24 | <0.0001 | <0.0001 |
| IL6 | 415.9 ± 506.5 | 223.9 | 1029.1 ± 2035.0 | 297.5 | 1.68 | 0.093 | 0.201 |
| IL8 | 14566.5 ± 6607.5 | 13105.9 | 11351.0 ± 4887.4 | 10943.4 | −2.69 | 0.007 | 0.022 |
| IL9 | 22.3 ± 41.6 | 5.7 | 50.7 ± 108.8 | 16.1 | 2.68 | 0.007 | 0.021 |
| IL10 | 13.7 ± 59.3 | 3.0 | 35.5 ± 228.1 | 5.0 | 2.12 | 0.034 | 0.080 |
| IL13 | 49.1 ± 96.6 | 19.5 | 303.3 ± 523.0 | 91.8 | 5.30 | <0.0001 | <0.0001 |
| IL17A | 11.7 ± 18.2 | 7.7 | 45.0 ± 208.4 | 12.4 | 2.25 | 0.024 | 0.062 |
| IL17F | 31.0 ± 81.9 | 5.8 | 14.5 ± 27.4 | 5.8 | −0.59 | 0.553 | 0.653 |
| IL23 | 11.2 ± 19.2 | 7.1 | 12.9 ± 20.9 | 9.4 | 1.29 | 0.197 | 0.394 |
| IL33 | 2845.5 ± 5308.2 | 1155.3 | 6486.7 ± 20938.6 | 1326.3 | 0.60 | 0.550 | 0.681 |
| TNFα | 17.3 ± 22.7 | 8.5 | 96.4 ± 418.7 | 11.0 | 0.97 | 0.334 | 0.511 |
| VEGF-A | 4957.6 ± 1798.2 | 4610.6 | 3985.2 ± 1407.2 | 3784.8 | −3.53 | 0.0004 | 0.0018 |
LEGEND: All reported protein concentrations are expressed in picograms per milliliter (pg./mL). OC, olfactory cleft; CRS, chronic rhinosinusitis; CRSsNP, CRS without nasal polyposis; CRSsNP, CRS without nasal polyposis; N, sample size; SD, standard deviation; MWU, Mann-Whitney U test statistic; CCL, chemokine (C-C motif) ligand; CXCL, C-X-X motif chemokine; EGF, epidermal growth factor; bFGF, basic fibroblast growth factor; IgE, immunoglobulin E; IL, interleukin; TNFα, tumor necrosis factor alpha; VEGF-A, vascular endothelial growth factor-A. Q-values, Benjamini-Hochberg p-value adjustment using a 5% false discovery rate.
In addition to examining differences in absolute protein levels between CRS and control cohorts, we examined the proportion of individual patients with CRS having abnormal levels for each protein (Table 4). This enabled us to determine what percentage of patients may be driving any variations in absolute protein levels. We found that CCL2, CCL3, IgE, IL5, IL9 and IL13 were elevated in a significantly higher proportion of patients with CRSwNP compared to those with CRSsNP.
Table 4:
Proportion of patients with abnormal cytokine levels
| Cytokine: | Percent of patients in each group having abnormal protein expression N (%) |
Test statistic: CRSsNP vs. CRSwNP | 95% CI | Unadjusted p-values |
q-values | ||
|---|---|---|---|---|---|---|---|
| All subjects with CRS | CRSsNP | CRSwNP | |||||
| CCL2 | 22 (14.6%) | 2 (3.1%) | 20 (23.0%) | OR=9.25 | 2.08–41.23 | 0.0035 | 0.022 |
| CCL3 | 80 (53.0%) | 23 (35.9%) | 57 (65.5%) | OR=3.39 | 1.72–6.66 | 0.0004 | 0.003 |
| CCL5 | 50 (33.1%) | 22 (34.4%) | 28 (32.2%) | OR=0.91 | 0.46–1.80 | 0.777 | 0.883 |
| CCL11 | 21 (13.9%) | 9 (14.1%) | 12 (13.8%) | OR=0.98 | 0.39–2.48 | 0.962 | >0.990 |
| CCL20 | 68 (45.0%) | 29 (45.3%) | 39 (44.8%) | OR=0.98 | 0.51–1.88 | 0.952 | >0.990 |
| CXCL1 | 106 (70.2%) | 47 (73.4%) | 59 (67.8%) | OR=0.76 | 0.37–1.56 | 0.456 | 0.713 |
| CXCL5 | 29 (19.2%) | 13 (20.3%) | 16 (18.4%) | OR=0.88 | 0.39–2.00 | 0.767 | 0.913 |
| CXCL9 | 26 (17.2%) | 11 (17.2%) | 15 (17.2%) | OR=1.00 | 0.43–2.36 | 0.993 | 0.993 |
| CXCL11 | 21 (13.9%) | 12 (18.8%) | 9 (10.3%) | OR=0.50 | 0.20–1.27 | 0.140 | 0.438 |
| EGF | 21 (13.9%) | 10 (15.6%) | 11 (12.6%) | OR=0.78 | 0.31–1.97 | 0.601 | 0.791 |
| bFGF | 70 (46.4%) | 28 (43.8%) | 42 (48.3%) | OR=1.20 | 0.63–2.30 | 0.582 | 0.808 |
| IgE | 46 (32.6%) | 14 (23.3% | 32 (39.5%) | OR=2.15 | 1.02–4.52 | 0.043 | 0.179 |
| IL2 | 25 (18.5%) | 11 (20.4%) | 14 (17.3%) | OR=0.82 | 0.34–1.97 | 0.651 | 0.814 |
| IL4 | 10 (7.4%) | 3 (5.6%) | 7 (8.6%) | OR=1.61 | 0.40–6.51 | 0.502 | 0.738 |
| IL5 | 57 (42.2%) | 7 (13.0%) | 50 (61.7%) | OR=10.83 | 4.35–26.95 | <0.0001 | <0.0001 |
| IL6 | 31 (23.0%) | 10 (18.5%) | 21 (25.9%) | OR=1.54 | 0.66–3.59 | 0.316 | 0.527 |
| IL8 | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | ---- | ---- | ---- | ---- |
| IL9 | 30 (22.2%) | 7 (13.0%) | 23 (28.4%) | OR=2.66 | 1.05–6.74 | 0.035 | 0.175 |
| IL10 | 76 (56.3%) | 26 (48.1%) | 50 (61.7%) | OR=1.74 | 0.87–3.49 | 0.119 | 0.425 |
| IL13 | 56 (41.5%) | 8 (14.8%) | 48 (59.3%) | OR=8.36 | 3.50–20.00 | <0.0001 | 0.0001 |
| IL17A | 34 (25.2%) | 10 (18.5%) | 24 (29.6%) | OR=1.85 | 0.80–4.27 | 0.145 | 0.403 |
| IL17F | 10 (6.6%) | 6 (11.1%) | 4 (4.9%) | OR=0.42 | 0.11–1.55 | 0.180 | 0.450 |
| IL23 | 62 (45.9%) | 21 (38.9%) | 41 (50.6%) | OR=1.61 | 0.80–3.24 | 0.180 | 0.411 |
| IL33 | 31 (23.0%) | 10 (18.5%) | 21 (25.9%) | OR=1.54 | 0.66–3.59 | 0.316 | 0.564 |
| TNFα | 70 (51.9%) | 25 (46.3%) | 45 (55.6%) | OR=1.45 | 0.73–2.90 | 0.292 | 0.562 |
| VEGF-A | 22 (14.6%) | 12 (18.8%) | 10 (11.5%) | OR=0.56 | 0.23–1.40 | 0.212 | 0.442 |
Legend: N, sample size; CRS, chronic rhinosinusitis; CRSsNP, chronic rhinosinusitis without nasal polyposis; CRSwNP, chronic rhinosinusitis with nasal polyposis; CI, confidence interval of the odds ratio; OR, odds ratio; CCL, chemokine (C-C motif) ligand; CXCL, C-X-X motif chemokine; EGF, epidermal growth factor; bFGF, basic fibroblast growth factor; IgE, immunoglobulin E; IL, interleukin; TNFα, tumor necrosis factor alpha; VEGF-A, vascular endothelial growth factor-A. Q-values, Benjamini-Hochberg p-value adjustment using a 5% false discovery rate.
For example, the well-known type-2 cytokine, IL5, was elevated in 13% of patients with CRSsNP and 61% of patients with CRSwNP (OR=10.83), and absolute IL5 levels in the CRSwNP cohort are significantly elevated compared to CRSsNP (Table 2). Other type-2 mediators, including IL13 and IgE, followed a similar pattern. On the other hand, CXCL1 is an example of a cytokine that is increased in the overall CRS cohort, both in absolute values (Table 2), as well as prevalence of patients with elevated values. However, levels of CXCL1 do not vary between phenotypes, thus it appears to be a non-specific inflammatory mediator that is elevated in approximately 70% of patients with CRS regardless of nasal polyposis. IL8 had decreased levels in the overall CRS cohort compared to controls (Table 2) and decreased levels in CRSwNP compared to CRSsNP (Table 3). However, given the wide normative reference range in controls, none of the patients with CRS were deemed to have abnormal values (Table 4). Thus there appears to be variability in IL8 levels based upon CRS phenotype, but these differences lie within the normal range. Additional normative cytokine expression measures in case subjects with CRS, with and without nasal polyposis, are provided for both patients with normal olfactory function (TDI total score≥31.0; Supplementary Table 1) and abnormal olfaction (TDI total score < 31.0; Supplementary Table 2).
Bivariate Associations between Olfactory Function and OC Mucus Protein Concentrations
A summary of the primary correlation analyses is provided in Supplemental Table 3. Adjusted, bivariate correlations between OC mucus protein concentrations and Sniffin’ Sticks total olfaction scores are presented in Table 5. In all subjects with CRS (n=151), significant, weak to moderate correlations between higher protein concentrations and worse olfactory function were identified for CCL2, CCL3, IgE, IL5, IL6, IL9, IL10, and IL13. Conversely, higher concentrations of CXCL11 and VEGF-A were associated with higher olfactory function scores when considering all cases. Before p-value adjustment, significant correlations within case subjects with CRSwNP were identified between Sniffin’ Stick total scores and CCL2, CCL3, CCL5, IL6, IL0, CXCL5, and VEGF-A. The correlation and unadjusted p-values demonstrate some interesting significant correlations to total olfaction score among several proteins in all subjects with CRS (CCL2, IL5, CCL3, IL13, IgE, IL6, IL-10, 1l-9, VEGF-A and CXCL5) and in CRSwNP (CCL2, CCL3, IL6, IL-10, CXCL5, and CCL5). After adjustment for multiplicity, no statistically significant correlations were identified for case subjects with CRSwNP. To minimize multiple statistical assessments, we focused subsequent analyses upon the 12 cytokines which had any significant findings (Table 5).
Table 5:
Spearman’s bivariate correlation coefficients between olfactory cleft mucus protein concentrations and Sniffin’ Sticks total scores
| Correlations with Sniffin’ Sticks total olfaction score | ||||||
|---|---|---|---|---|---|---|
| All Case Subjects with CRS | CRSsNP | CRSwNP | ||||
| R (unadj. p-value) | q-value | R (unadj. p-value) | q-value | R (unadj. p-value) | q-value | |
| CCL2 | −0.388 (<0.001) | <0.0001 | −0.213 (0.094) | >0.990 | −0.258 (0.016) | 0.208 |
| IL5 | −0.378 (<0.001) | <0.0001 | −0.073 (0.601) | >0.990 | −0.167 (0.136) | 0.321 |
| CCL3 | −0.353 (<0.001) | <0.0001 | −0.081 (0.530) | >0.990 | −0.223 (0.038) | 0.198 |
| IL13 | −0.336 (<0.001) | 0.0005 | −0.074 (0.598) | >0.990 | −0.197 (0.079) | 0.257 |
| IgE | −0.295 (<0.001) | 0.002 | 0.030 (0.823) | >0.990 | −0.187 (0.096) | 0.277 |
| IL6 | −0.250 (0.004) | 0.015 | −0.045 (0.747) | >0.990 | −0.281 (0.011) | 0.286 |
| IL10 | −0.244 (0.004) | 0.016 | −0.042 (0.764) | >0.990 | −0.242 (0.030) | 0.260 |
| IL9 | −0.233 (0.007) | 0.019 | −0.123 (0.381) | >0.990 | −0.080 (0.477) | 0.775 |
| VEGF-A | 0.229 (0.005) | 0.016 | −0.025 (0.846) | >0.990 | 0.211 (0.050) | 0.186 |
| CXCL11 | 0.213 (0.009) | 0.023 | 0.050 (0.694) | >0.990 | 0.179 (0.098) | 0.255 |
| CXCL5 | 0.144 (0.079) | 0.186 | 0.198 (0.120) | >0.990 | 0.225 (0.036) | 0.234 |
| TNFα | −0.140 (0.106) | 0.212 | −0.015 (0.917) | >0.990 | −0.156 (0.164) | 0.355 |
| CXCL9 | 0.133 (0.105) | 0.228 | 0.078 (0.545) | >0.990 | 0.142 (0.190) | 0.353 |
| bFGF | −0.128 (0.119) | 0.221 | −0.071 (0.579) | >0.990 | −0.143 (0.186) | 0.372 |
| CCL5 | −0.112 (0.173) | 0.300 | −0.035 (0.784) | >0.990 | −0.212 (0.049) | 0.212 |
| EGF | 0.090 (0.275) | 0.447 | −0.005 (0.972) | >0.990 | 0.094 (0.384) | 0.666 |
| IL17A | −0.085 (0.331) | 0.506 | −0.066 (0.637) | >0.990 | 0.067 (0.554) | 0.800 |
| CXCL1 | 0.065 (0.432) | 0.624 | −0.095 (0.459) | >0.990 | 0.068 (0.534) | 0.817 |
| IL8 | 0.063 (0.468) | 0.640 | −0.143 (0.308) | >0.990 | −0.010 (0.928) | 0.965 |
| IL4 | −0.053 (0.545) | 0.709 | −0.143 (0.308) | >0.990 | 0.045 (0.688) | 0.852 |
| IL2 | −0.049 (0.575) | 0.712 | −0.060 (0.671) | >0.990 | −0.011 (0.924) | >0.999 |
| IL23 | −0.044 (0.617) | 0.729 | 0.001 (0.998) | >0.990 | 0.050 (0.655) | 0.852 |
| IL33 | −0.044 (0.618) | 0.698 | −0.015 (0.915) | >0.990 | −0.007 (0.953) | 0.953 |
| CCL20 | −0.031 (0.702) | 0.761 | −0.031 (0.809) | >0.990 | −0.034 (0.753) | 0.851 |
| CCL11 | −0.004 (0.963) | >0.990 | 0.019 (0.881) | >0.990 | −0.057 (0.599) | 0.820 |
| IL17F | 0.004 (0.962) | 0.962 | −0.014 (0.922) | >0.990 | −0.044 (0.695) | 0.821 |
LEGEND: CRS, chronic rhinosinusitis; CRSsNP, CRS without nasal polyposis; CRSwNP, CRS with nasal polyposis; R, Spearman’s rank correlation coefficient; CCL, chemokine (C-C motif) ligand; CXCL, C-X-X motif chemokine; EGF, epidermal growth factor; bFGF, basic fibroblast growth factor; IgE, immunoglobulin E; IL, interleukin; TNFα, tumor necrosis factor alpha; VEGF-A, vascular endothelial growth factor-A. R values are listed in decreasing magnitude for all case subjects with CRS with associated adjusted p-values. Q-values, Benjamini-Hochberg p-value adjustment using a 5% false discovery rate.
Association of abnormal cytokine levels with clinical olfactory function
Given significant correlations between protein levels and total Sniffin Sticks scores, we then compared total TDI scores between patients with normal and abnormal cytokine values to associate the cytokine values with olfactory function. Table 6 demonstrates that patients with abnormal levels of CCL2, CCL3, IL5, IL10 and IL13 all suffer from TDI deficits of approximately 5 points or greater. This results in expected differences in classification of these patients into functional olfactory diagnostic categories. Some variability can be seen in which cytokines are associated with olfactory testing category. For example, over half of anosmics display elevated CCL3, IL5, IL10 or IL13. While CCL2 and IL6 associate with olfactory categorization, it can be seen that these cytokines are abnormal in 1/3 or fewer of anosmic patients.
Table 6:
Mean Sniffin’ Sticks TDI scores for patients with CRS with normal and abnormal cytokines levels and prevalence of patients with abnormal cytokine levels within each olfactory category
| Cytokine: | Total TDI | Prevalence of abnormal protein level for each olfactory category | Test statistic | Unadjusted p-values |
q-values | |||
|---|---|---|---|---|---|---|---|---|
| Normal protein level | Abnormal protein level | Normosmia (N=38): N (%) |
Hyposmia (N=63): N (%) |
Anosmia (N=49): N (%) |
||||
| CCL2 | 23.2 ± 9.0 | 15.6 ± 8.5 | 3 (7.9%) | 5 (7.9%) | 14 (28.6%) | χ2=11.24 | 0.004 | 0.012 |
| CCL3 | 25.2 ± 8.7 | 19.3 ± 9.0 | 13 (34.2%) | 30 (47.6%) | 36 (73.5%) | χ2=14.34 | 0.001 | 0.006 |
| CCL5 | 22.3 ± 9.4 | 21.8 ± 9.4 | 10 (26.3%) | 22 (34.9%) | 17 (34.7%) | χ2=0.93 | 0.627 | 0.752 |
| CXCL5 | 21.5 ± 9.4 | 24.4 ± 8.6 | 9 (23.7%) | 12 (19.0%) | 8 (16.3%) | χ2=0.75 | 0.688 | 0.750 |
| CXCL11 | 21.8 ± 9.3 | 24.2 ± 9.1 | 6 (15.8%) | 9 (14.3%) | 6 (12.2%) | χ2=0.23 | 0.891 | 0.891 |
| IgE | 22.6 ± 9.1 | 20.8 ± 9.8 | 10 (27.8%) | 16 (27.6%) | 20 (43.5%) | χ2=3.50 | 0.173 | 0.296 |
| IL5 | 24.4 ± 8.8 | 17.7 ± 9.0 | 7 (21.2%) | 19 (35.8%) | 31 (64.6%) | χ2=16.65 | <0.001 | 0.0001 |
| IL6 | 22.0 ± 9.1 | 20.3 ± 10.7 | 10 (30.3%) | 5 (9.4%) | 16 (33.3%) | χ2=9.36 | 0.009 | 0.022 |
| IL9 | 22.0 ± 9.5 | 20.0 ± 9.5 | 6 (18.2%) | 11 (20.8%) | 13 (27.1%) | χ2=1.03 | 0.599 | 0.798 |
| IL10 | 24.1 ± 8.7 | 19.6 ± 9.6 | 15 (45.5%) | 26 (49.1%) | 34 (70.8%) | χ2=6.81 | 0.033 | 0.066 |
| IL13 | 24.3 ± 8.7 | 17.7 ± 9.2 | 8 (24.2%) | 18 (34.0%) | 30 (62.5%) | χ2=13.98 | 0.001 | 0.004 |
| VEGF-A | 21.9 ± 9.2 | 23.3 ± 10.1 | 7 (18.4%) | 7 (11.1%) | 8 (16.3%) | χ2=1.17 | 0.556 | 0.834 |
Legend: TDI, Sniffin’ Stick TDI total score; N, sample size; χ2, chi-square test; CCL, chemokine (C-C motif) ligand; CXCL, C-X-X motif chemokine; IgE, immunoglobulin E; IL, interleukin; VEGF-A, vascular endothelial growth factor-A; χ2, chi-squared test statistic. Q-values, Benjamini-Hochberg p-value adjustment using a 5% false discovery rate.
Correlation between Clinical Measures of Disease Severity and OC Mucus Protein Concentrations
The relationships between individual OC mucus protein concentrations and LM scores are summarized for all case subjects with CRS as well as subgroups of CRSsNP and CRSwNP (Table 7). CT imaging was not collected on healthy, control subjects. Greater overall opacification on CT scores for case subjects was found to correlate with higher concentrations of CCL2, CCL3, IgE, IL4, IL5, IL6, IL9, IL10, IL13, IL17A, and IL33 and inversely with IL8 and VEGF-A without adjustment for family-wise comparisons. After adjustment, similar correlations remained with the exception of IL6 and IL9. Prior to adjustment, significant findings were differential for CRSsNP and CRSwNP subgroups: in CRSsNP, CCL3, IL5, and IL13 demonstrated weak to moderate significant correlation with higher LM CT scores (p<0.050) whereas CCL2 and IL10 demonstrated weak but significant correlation with LM CT scores in CRSwNP (p<0.050). These associations were non-significant after adjustment using the Benjamini-Hochberg procedure.
Table 7:
Spearman’s bivariate correlation coefficients between olfactory cleft mucus protein concentrations, Lund-Mackay CT scoring, and olfactory cleft opacification.
| Correlation with Lund-Mackay CT scores | |||
|---|---|---|---|
| All Case Subjects with CRS | CRSsNP | CRSwNP | |
| R (q-value) | R (q-value) | R (q-value) | |
| CCL2 | 0.375 (<0.0001) | 0.096 (0.495) | 0.240 (0.132) |
| CCL3 | 0.350 (<0.0001) | 0.275 (0.116) | 0.063 (0.695) |
| CCL5 | 0.031 (0.780) | −0.158 (0.430) | 0.152 (0.437) |
| CXCL5 | −0.031 (0.718) | −0.131 (0.525) | −0.068 (0.733) |
| CXCL11 | −0.162 (0.065) | 0.122 (0.453) | −0.045 (0.757) |
| IgE | 0.338 (0.0002) | 0.269 (0.096) | −0.010 (0.931) |
| IL5 | 0.489 (<0.0001) | 0.314 (0.132) | 0.139 (0.411) |
| IL6 | 0.186 (0.055) | 0.044 (0.752) | 0.155 (0.380) |
| IL9 | 0.178 (0.061) | 0.141 (0.470) | −0.129 (0.413) |
| IL10 | 0.340 (0.0002) | 0.130 (0.422) | 0.294 (0.144) |
| IL13 | 0.455 (<0.0001) | 0.317 (0.252) | 0.210 (0.222) |
| VEGF-A | −0.385 (<0.0001) | −0.264 (0.111) | −0.258 (0.132) |
| Correlation with Average Percent Opacification of Olfactory Cleft | |||
| All Case Subjects with CRS | CRSsNP | CRSwNP | |
| R (q-value) | R (q-value) | R (q-value) | |
| CCL2 | 0.382 (<0.0001) | 0.077 (0.707) | 0.268 (0.156) |
| CCL3 | 0.280 (0.003) | −0.129 (0.542) | 0.173 (0.264) |
| CCL5 | 0.158 (0.092) | 0.139 (0.555) | 0.244 (0.129) |
| CXCL5 | −0.117 (0.202) | −0.333 (0.192) | −0.028 (0.897) |
| CXCL11 | −0.196 (0.041) | 0.042 (0.837) | −0.022 (0.859) |
| IgE | 0.331 (0.0009) | 0.088 (0.733) | 0.080 (0.709) |
| IL5 | 0.543 (<0.0001) | 0.310 (0.240) | 0.211 (0.184) |
| IL6 | 0.172 (0.088) | −0.024 (0.878) | 0.272 (0.116) |
| IL9 | 0.314 (0.002) | 0.236 (0.369) | 0.071 (0.688) |
| IL10 | 0.332 (0.0008) | 0.241 (0.460) | 0.247 (0.115) |
| IL13 | 0.416 (<0.0001) | 0.215 (0.386) | 0.142 (0.392) |
| VEGF-A | −0.390 (<0.0001) | −.0.147 (0.594) | −0.292 (0.180) |
LEGEND: CRS, chronic rhinosinusitis; CRSsNP, CRS without nasal polyposis; CRSwNP, CRS with nasal polyposis; R, Spearman’s rank correlation coefficient; CT, computed tomography; CCL, chemokine (C-C motif) ligand; CXCL, C-X-X motif chemokine; IgE, immunoglobulin E; IL, interleukin; VEGF-A, vascular endothelial growth factor-A. R values are listed in decreasing magnitude for all case subjects with CRS. Q-values, Benjamini-Hochberg p-value adjustment using a 5% false discovery rate.
Comparable findings were identified after evaluating unadjusted, bivariate correlations between OC protein concentrations and total percentages of volumetric opacification on CT scans for case subjects (Table 7). The highest magnitude of correlation between the total percent of OC opacification was found to be with IL5 (r=0.543; q<0.0001) while other moderate correlations were noted with IgE, IL10 and IL13.
DISCUSSION
It is now well accepted that CRS represents a heterogeneous collection of underlying pathobiology and inflammatory pathways, often referred to as endotypes. As our understanding of the complexities of the inflammatory processes evolves, so does our understanding of the variety of mechanisms that may be associated with CRS-related OD. Along with this study, there is a growing body of literature that supports heterogeneous etiologies for OD in CRS.1,6,20
The current study addresses the cytokine milieu of the OC and relates that to olfactory function with the aid of a control group. This cytokine milieu is but one factor that may be associated with olfactory function. There has been relatively little attention focused on assessing individual mechanisms through which OD occurs in cohorts with CRS. Several broad mechanisms have been suggested and there are potential mechanisms that would not be captured by OC cytokines, like nasal cavity polyps blocking airflow or age-related central nervous system changes. 1,6,20
While several endotyping studies have been reported on patients with CRS, all have been limited by two major pitfalls; the use of sinonasal tissue or a focus upon only type-1/type-2 cytokines. Studies utilizing tissue significantly limit themselves to only the most severely affected patients having surgery and possibly exclude those who may respond to medical management. Surgical tissue studies also exclude patients with comorbidities (e.g. elderly patients) that make them less suitable for elective surgery. We have collected mucous as a non-invasive, atraumatic sample. Studies utilizing nasal mucus for endotyping have largely focused on type-1 and type-2 cytokines which do not fully represent the complexity and broad range of cell types involved in the pathogenesis of CRS.21,22 Herein, we evaluated a diverse panel of cytokines, chemokines, growth factors and neuroinflammatory mediators including those considered to be type 1,2 and 3 inflammatory modulators to aid in a better understanding of their biological role in disease severity and olfactory loss. While the mucus is collected from the OC, we are unable to determine its precise origin and it is possible it originates in the OC, that it is transported from nasal mucosa to the OC, or some combination of both. As the use of nasal mucus for biomarker research expands, development of new nasal mucus biomarkers will need to be rigorously tested and normative ranges established before this technology can be used clinically.
This study confirmed that many inflammatory proteins vary both by disease phenotype and also in their association with olfaction. Using normative ranges established from a control population for each protein, in this study we are able to overcome prior limitations from group comparisons in order to provide personalized data. This allowed us to determine not only differences in means/medians between CRS cohorts, but to determine what proportion of patients in a given cohort was driving these differences with implications for personalized treatment approaches.
It is not surprising that classic type-2 inflammatory mediators are increased in CRS, especially driven by the nasal polyp group. However, different patterns can be observed among mediators. For example, IL13 is elevated in CRSwNP compared to CRSsNP and this is driven in large part by approximately 60% of CRSwNP patients who have abnormally elevated levels of this cytokine. This has direct clinical relevance, especially in the era of targeted biologic therapies.7, 23, 24 Prior studies of dupilumab, an IL4/13 antagonist, report a mean improvement in NP score of approximately 2 points on an 8 point scale for the entire cohort.25 In responder analyses, 45% of patients improve by 2 points or more whereas nearly half of all patients do not improve or only improve by 1 point. This approximates the 60% of patients with elevated IL13 levels in the present study. Additionally, it is interesting that IL4 levels were not elevated in our study, similar to other reports,21 thus dupilumab effects may be via IL13 mediated mechanisms. The precise mechanism of dupilumab with respect to IL4R blockade is unclear but may also be related to mast cell function or other upstream effects. The question of dual inhibition (anti-IL14 and anti-IL13) vs anti-IL13 alone will be debated as more specific inhibitors come from development and clinical trials. The increase in IL13 herein suggests that dual inhibition may not be required for the olfactory benefit, though this deserves further study.
Other type-2 mediators also play a role with IL-5 having abnormal expression in 61.7% of CRSwNP patients, similar to IL13. IgE was elevated however in only 40% of CRSwNP patients. Recent clinical trial data suggest that roughly 40% of patients treated with omalizumab have improvement in NP scores of 2 or greater.26 Again, this suggests that identifying patients with abnormally elevated biomarkers may be useful for predicting individual clinical response to targeted monoclonal antibody therapies. Type 3 inflammatory mediators including IL-17A, IL-23, IL-6, demonstrated increased concentrations in case subjects with CRS compared to control subjects. When comparing CRSwNP to those without polyps, no difference was observed with IL-6 or IL-23 while an increase in IL-17 was observed in CRSwNP. IL-6 had low to moderate negative correlation with total olfaction score in CRSwNP while IL-17A and IL-23 had no significant correlation with olfactory testing in case subjects with CRS.
Novel information regarding the inflammatory profile of CRSsNP was also gained from this study. Prior studies have suggested that there is overlap between inflammatory profiles in CRSwNP and CRSsNP. IL5 and IL13 were elevated in approximately 15% of CRSsNP patients, suggesting that there is a substantial number of patients with type-2 inflammation despite the absence of gross polyposis. This group, without polyposis, may also benefit from targeted biologics currently indicated only in patients with polyposis. Taken together, these mucus cytokine data demonstrate the potential for different targeted therapies to be directed to specified groups of CRSwNP and even some CRSsNP patients. Ultimately, additional prospective trials will be necessary to answer this important clinical question. The data from this study could inform study design of future clinical trials since it provides normative and abnormal values with means, standard deviations and ranges.
With respect to olfactory testing, the OC protein concentrations correlate with psychophysical testing in CRS. In patients with CRS, this was driven primarily by type-2 inflammatory mediators including CCL2 and 3, IgE, IL5, and IL13. Even when stratified for polyp status, these differences persist. This suggests that the OC cytokine profile is not simply a marker of polyp status and that the pathophysiology of OD in CRS includes localized tissue inflammation in addition to obstruction of nasal airflow. CCL2 is interesting in that it was significantly correlated with Sniffin’ Sticks in CRS with and without NP. In a previous study of individuals with olfactory loss not caused by CRS, CCL2 was also implicated.6 Of note, CCL2 was highly correlated with TDI in a prior CRS cohort (R=0.5, p<0.001)1 and it also correlated with TDI in a prior control cohort (R=0.5; p=0.004).6 One possibility is that CCL2 is a marker of a non-CRS etiology for olfactory loss. Further study is warranted as CCL2 blocking antibodies are currently being investigated in the management of other disease processes.27
When exploring associations between OC protein concentrations and psychophysical testing of olfactory function (Table 5), we have presented correlations with unadjusted p-values and Q values which are Benjamini-Hochberg (BH) p-value adjustments using a 5% false discovery rate. While the correlation and unadjusted p-values demonstrate some interesting correlations to total olfaction score among several proteins in all subjects with CRS (CCL2, IL5, CCL3, IL13, IgE, IL6, IL-10, 1l-9, VEGF-A and CXCL5) and in CRSwNP (CCL2, CCL3, IL6, IL-10, CXCL5, and CCL5), these correlations lose statistical significance after BH adjustments. This is not surprising as it would take a much larger sample size to be able to show a difference in specific subgroups after BH multiple comparisons adjustments since the correlations, as anticipated, are not high. Multiple comparisons adjustments have a place in the analytic spectrum but they should not be used to solely interpret the results of this study.
Our data demonstrate some interesting findings with regard to OC cytokine concentration and LM scores. While several cytokines correlate with LM scoring across all CRS subjects, there is some differential expression between those with and without nasal polyps. In our prior study1 and in this cohort, OC IL5 and IL13 concentrations correlate with LM scores in CRSsNP. This supports the likelihood for endotype heterogeneity within clinical phenotypes, with particular importance for type-2 mediators. It is possible that these patients are more likely to have diffuse inflammation which is reflected in higher opacification of the paranasal sinuses and type-2 inflammatory cytokines in the OC.
The questions of whether endotypes change over time within an individual or if cytokine levels vary with olfactory function in a given individual are among those we have considered. There is speculation that this could occur over time via natural history of the disease, that mucus cytokines may not represent underlying endotype, or that cytokines could be altered based on various treatments; however, data exploring this question is quite limited at present.28 Our study was designed to be cross-sectional and thus we do not have longitudinal data points to examine stability or reflection of underlying endotype. Thus caution should be exercised when using individual cytokines obtained at a single point in time to define patient endotypes and subsequent therapies. This would be an important area of study in the future and would probably best be done in the confines of individual placebo controlled clinical trials, where isolated treatments can be rigorously controlled and assessments performed at pre-determined time points.
A number of issues should be considered as these data are analyzed. This cross-sectional, controlled cohort study provides data prospectively collected from multiple centers for the specific purpose of understanding the OC inflammatory milieu and to associate that, where possible, with psychophysical testing for olfaction and other measures of disease and disease severity. Given the intricate and multifactorial nature of CRS-OD, compounded by inherent challenges of human cohort studies, it is no surprise that putative factors and their relative importance have been inconsistently characterized across studies. The great challenges for the application of precision medicine in olfactory disorders are the idiosyncrasies, nuances, and numerous possible interrelated factors. Much work remains to be done in this field and without accounting for all factors that might affect olfaction, there is risk of reporting spurious results. Multiple proteins were assayed and include a range of type 1,2, and 3 inflammatory proteins. It is possible that other unrelated disease mechanisms are at play and/or that the proteins measured are simply markers of inflammatory disease not specific to OD. We have stratified for polyposis to prevent over-interpretation of results for all CRS based simply on polyp related inflammatory processes. The issue of multiplicity associated with a large number of outcome variables within a relatively small sample must be considered. We have utilized a false discovery ratio (FDR) to control the number of false positive and avoid further type-I errors. To that end, we elected to include Benjamini-Hochberg adjusted p-values (“q-values”) in Tables 2–6 using a highly conservative 5% FDR. Of note, patients were permitted to be on baseline maintenance medical therapy for CRS with the exception of oral corticosteroids at the time of sampling. This reflects the clinical situation where we expect endotyping would be employed in the future for clinical decision-making. As we considered a control group during study design, we selected a control group without sinonasal disease or medications. For future studies, there is room to select different control groups which would allow for some interesting comparisons. The panel of proteins assayed was determined mainly by the commercially available bead array. Future studies should include a global proteomics approach to define the array of proteins even beyond inflammatory proteins such as those implicated in odorant transport and metabolization. Finally, for this study, samples were only collected and analyzed from the OC and we are not able to comment on the cytokine profile of the middle meatus which will be an important consideration for future studies.
CONCLUSIONS
A growing body of literature supports heterogeneous etiologies for CRS-OD. We have examined a diverse panel of OC cytokines, chemokines, growth factors and neuroinflammatory mediators to aid in understanding their association with disease phenotype, endotype, and olfactory function. This study confirmed that many inflammatory proteins vary both by disease phenotype and also in their association with olfactory function. Classic type-2 inflammatory mediators are increased in CRS, especially driven by the CRSwNP group. However, only 40% to 60% of CRSwNP patients have elevation of specific type-2 mediators and would be expected to respond to current targeted monoclonal antibody therapy. Conversely, a substantial proportion (approximately 15%) of CRSsNP patients, often eliminated from consideration for monoclonal antibody therapy, have elevated type-2 inflammatory mediators. Further research is necessary to understand the complex roles OC mucous inflammatory proteins might play in defining endotype and in impacting CRS-related OD.
Supplementary Material
Funding source(s):
TLS, RJS, ZMS, JCM, V.R.R., J.A.A. and J.K.M were supported by a grant from the National Institute on Deafness and Other Communication Disorders (NIDCD), one of the National Institutes of Health, Bethesda, MD (R01 DC005805; PIs: TLS and ZMS). This funding organization did not contribute to the design or conduct of this study; preparation, review, approval or decision to submit this manuscript for publication.
Footnotes
Potential Conflicts of Interest: None of the following are affiliated with this study. Z.M.S.: Olympus, OptiNose, Regeneron, Healthy Humming, and Novartis (consultant). J.A.A.: Medtronic and GlycoMira Therapeutics (consultant). R.J.S.: OptiNose, Olympus, Stryker, Regeneron, and Healthy Humming (consultant). V.R.R.: OptiNose and Medtronic, Inc, (consultant). D.M.B: Medtronic (consultant).
REFERENCES
- 1.Soler ZM, Yoo F, Schlosser RJ, et al. Correlation of mucus inflammatory proteins and olfaction in chronic rhinosinusitis. Int Forum Allergy Rhinol 2020; 10(3): 343–355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hauser LJ, Chandra RK, Li P, Turner JH. Role of tissue eosinophils in chronic rhinosinusitis-associated olfactory loss. Int Forum Allergy Rhinol 2017; 7(10):957–962. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lavin J, Min JY, Lidder AK, et al. Superior turbinate eosinophilia correlates with olfactory deficit in chronic rhinosinusitis patients. Laryngoscope. 2017; 127(10):2210–2218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Holbrook EH, Rebeiz L, Schwob JE. Office-based olfactory mucosa biopsies. Int Forum Allergy Rhinol 2016; 6(6):646–653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lanza DC, Deems DA, Doty RL, et al. The effect of human olfactory biopsy on olfaction: a preliminary report. Laryngoscope. 1994; 104(7):837–840. [DOI] [PubMed] [Google Scholar]
- 6.Yoo F, Soler ZM, Mulligan JK, et al. Olfactory cleft mucus proteins associated with olfactory dysfunction in a cohort without chronic rhinosinusitis. Int Forum Allergy Rhinol 2019: 9(10):1151–1158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Smith TL. Scientific abstracts for RhinoWorld 2019. Int Forum Allergy Rhinol 2019; 9(S2):S49–S124. [DOI] [PubMed] [Google Scholar]
- 8.Turner JH, Chandra RK, Li P, Bonnet K, Schlundt DG. Identification of clinically relevant chronic rhinosinusitis endotypes using cluster analysis of mucus cytokines. J Allergy Clin Immunol 2018; 141(5):1895–1897.e7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Rosenfeld RM, Piccirillo JF, Chandrasekhar SS, et al. Clinical practice guideline (update): adult sinusitis. Otolaryngol Head Neck Surg 2015; 152(2 Suppl):S1–S39. [DOI] [PubMed] [Google Scholar]
- 10.Lund VJ, Mackay IS. Staging in rhinosinusitis. Rhinology. 1993;31(4):183–184. [PubMed] [Google Scholar]
- 11.Loftus C, Schlosser RJ, Smith TL, et al. Olfactory cleft and sinus opacification differentially impact olfaction in chronic rhinosinusitis. Laryngoscope. 2019; October11. doi: 10.1002/lary.28332. Online ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kobal G, Hummel T, Sekinger B, Barz S, Roscher S, Wolf S. “Sniffin’ sticks”: screening of olfactory performance. Rhinology. 1996; 34(4):222–226. [PubMed] [Google Scholar]
- 13.Hummel T, Kobal G, Gudziol H, Mackay-Sim A. Normative data for the “Sniffin’ Sticks” including tests of odor identification, odor discrimination, and olfactory thresholds: an upgrade based on a group of more than 3,000 subjects. Eur Arch Otorhinolaryngol 2007; 264(3):237–243. [DOI] [PubMed] [Google Scholar]
- 14.Messtechnik B Sniffin’ Sticks Smell Test Instructions for Use Manual Wedel, Germany, 2016. [Google Scholar]
- 15.Schlosser RJ, Mulligan JK, Hyer JM, Karnezis TT, Gudis DA, Soler ZM. Mucous cytokine levels in chronic rhinosinusitis-associated olfactory loss. JAMA Otolaryngol Head Neck Surg 2016; 142(8):731–737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Harris EK, Boyd JC. Statistical Bases of Reference Values in Laboratory Medicine. New York: Marcel Dekker, Inc.; 1995. [Google Scholar]
- 17.Linnet K Two stage transformation systems for normalization of reference distributions evaluated. Clin Chem 1987; 33: 381–386. [PubMed] [Google Scholar]
- 18.Solberg HE, Grasbeck R. Reference values. Adv Clin Chem 1989; 27: 1–79. [DOI] [PubMed] [Google Scholar]
- 19.Young DS. Determination and validation of reference intervals. Arch Pathol Lab Med 1992; 116(7):704–709. [PubMed] [Google Scholar]
- 20.Mattos JL. Mechanisms and treatment of olfactory dysfunction in chronic rhinosinusitis. Ann Allergy Asthma Immunol 2020; 124(4):307–308 [DOI] [PubMed] [Google Scholar]
- 21.Morse JC, Shilts MH, Ely KA, et al. Patterns of olfactory dysfunction in chronic rhinosinusitis identified by hierarchical cluster analysis and machine learning algorithms. Int Forum Allergy Rhinol 2019; 9(3):255–264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wu J, Chandra RK, Li P, Hull BP, Turner JH. Olfactory and middle meatal cytokine levels correlate with olfactory function in chronic rhinosinusitis. Laryngoscope. 2018; 128(9):E304–E310. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Roland LT, Smith TL, Schlosser RJ, et al. Guidance for contemporary use of biologics in management of chronic rhinosinusitis with nasal polyps: discussion from a National Institutes of Health-sponsored workshop. Int Forum Allergy Rhinol 2020June2. doi: 10.1002/alr.22633. Online ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lal D, Borish L, Detwiller KY et al. A case for multidisciplinary management of chronic rhinosinusitis with nasal polyposis. Int Forum Allergy Rhinol. 2020; 10(7):795–797. [DOI] [PubMed] [Google Scholar]
- 25.Bachert C, Han JK, Desrosiers M, et al. Efficacy and safety of dupilumab in patients with severe chronic rhinosinusitis with nasal polyps (LIBERTY NP SINUS-24 and LIBERTY NP SINUS-52): results from two multicentre, randomised, double-blind, placebo-controlled, parallel-group phase 3 trials. Lancet 2019; 394:1638–1650. [DOI] [PubMed] [Google Scholar]
- 26.Gevaert P, Omachi TA, Corren J, et al. Efficacy and safety of omalizumab in nasal polyposis: two randomized phase III trials. J Allergy Clin Immunol 2020. June7;S0091-6749(20)30752-1. doi: 10.1016/j.jaci.2020.05.032. Online ahead of print. [DOI] [PubMed] [Google Scholar]
- 27.Gschwandtner M, Derler R, Midwood KS. More than just attractive: How CCL2 influences myeloid cell behavior beyond chemotaxis. Front Immunol 2019; 10:2759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Yancey KL, Li P, Huang LC, Sheng Q, Chandra RK, Chowdhury NI, Turner JH. Longitudinal stability of chronic rhinosinusitis endotypes. Clin Exp Allergy. 2019; 49(12):1637–1640. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.