Associations between inflammatory endotypes and clinical presentations in chronic rhinosinusitis

Abstract

Background

Chronic rhinosinusitis (CRS) is a heterogeneous disease characterized by mucosal inflammation in the nose and paranasal sinuses. Inflammation in CRS is also heterogeneous and is mainly characterized by type 2 (T2) inflammation but subsets of patients show type 1 (T1) and type 3 (T3) inflammation. Whether inflammatory endotypes are associated with clinical phenotypes has yet to be explored in detail.

Objective

To identify associations between inflammatory endotypes and clinical presentations in CRS.

Methods

We compared 121 non-polypoid CRS (CRSsNP) patients and 134 polypoid CRS (CRSwNP) patients and identified inflammatory endotypes using markers including IFN-γ (T1), eosinophil cationic protein (T2), Charcot-Leyden crystal galectin (T2) and IL-17A (T3). We collected clinical parameters from medical and surgical records and examined whether there were any associations between endotype and clinical features.

Results

The presence of nasal polyps (NPs), asthma comorbidity, smell loss and allergic mucin were significantly associated with the presence of T2 endotype in all CRS patients. The T1 endotype was significantly more common in females, and the presence of pus was significantly associated with T3 endotype in all CRS. We further analyzed these associations in CRSsNP and CRSwNP separately and found smell loss was still associated with T2 endotype and pus with the T3 endotype in both CRSsNP and CRSwNP. Importantly, CRS patients with T2 and T3 mixed endotype tended to have clinical presentations shared by both T2 and T3 endotypes.

Conclusions

Clinical presentations are directly associated with inflammatory endotypes in CRS. Identification of inflammatory endotypes may allow for more precise and personalized medical treatments in CRS.

INTRODUCTION

Chronic rhinosinusitis (CRS) is a heterogeneous disease that affects approximately 12.5% of Americans, is responsible for over 400,000 surgeries annually, produces significant morbidity and costs our health system an estimated $22–32 billion annually.^1–4 Most investigators accept a paradigm in which CRS is divided into the two main phenotypes: CRS with nasal polyps (CRSwNP) and CRS without nasal polyps (CRSsNP). In Western countries, CRSwNP is well known to be characterized by type 2 (T2) inflammation with pronounced eosinophilia and the presence of high levels of T2 cytokines, such as IL-5 and IL-13.^5–9 In contrast, CRSsNP is less well characterized despite 80% of all CRS patients having this phenotype.^{1, 2, 10} Recently, we comprehensively characterized inflammatory patterns by using IFN-γ as a type 1 (T1) marker, Charcot-Leyden crystal galectin (CLC) mRNA or eosinophil cationic protein (ECP) as T2 markers and IL-17A for a type 3 (T3) marker in CRS patients who had undergone surgery at our institution. We found that inflammation in CRSsNP is highly heterogeneous, much more so than what was observed in CRSwNP.⁷ Similar to other studies in the US and Europe, we reported that the most frequent inflammatory endotype in CRSsNP was T2, as opposed to T1 or T3.^{7–9, 11, 12} Additionally, on further analysis, we identified subsets of patients who had more than one type of inflammatory endotype, as characterized by concurrent elevations in T1, T2 and/or T3 markers.⁷ While these studies shed light onto the different molecular mechanisms underlying CRS, it remains unknown if (or how) this inflammation impacts the clinical presentation. Specifically, CRS patients experience diverse sinonasal and systemic symptoms,^13–15 but the relationship between a particular clinical phenotype and an inflammatory endotype is not well defined.

In the past decade, several groups have examined the association between endotypes and phenotypes in asthma. The Severe Asthma Research Program (SARP) and the Unbiased BIOmarkers in PREDiction of respiratory disease outcomes (U-BIOPRED) studies both identified asthma phenotypes and related them to inflammatory endotypes, including gene expression profiles, eosinophilia and neutrophilia.^16–21 In contrast to asthma, however, investigations into the association between endotype and phenotype of CRS remain scarce, except for the known association of T2 inflammation with nasal polyps (NPs) and asthma.^{8, 22} In this study, we hypothesized that certain CRS inflammatory endotypes are associated with specific clinical presentations. We therefore set out to define the endotype-phenotype associations in CRS by examining clinical characteristics of CRS patients evaluated in our previously published endotyping study.⁷

METHODS

Patients and tissue collection

We used mRNA and protein data from our published study, which included 255 CRS patients.⁷ All CRS patients were recruited from the Otolaryngology clinic and the Northwestern Sinus Center of Northwestern Medicine. All CRS patients met the criteria for CRS as defined by the International Consensus Statement on Allergy and Rhinology: Rhinosinusitis.² Patients with an established immunodeficiency, pregnancy, coagulation disorder or diagnosis of aspirin hypersensitivity, classic allergic fungal sinusitis, eosinophilic granulomatous polyangiitis (eGPA or Churg-Strauss syndrome) or cystic fibrosis were excluded from the study. Characteristics of subjects in this study are shown in Table 1. All subjects signed informed consent and the study was approved by the Institutional Review Board of Northwestern University Feinberg School of Medicine (IRB Project Number: STU00016917).

Table 1.

Clinical characteristics of patients with CRSsNP and CRSwNP

	CRSsNP % (n=121)	CRSwNP % (n=134)	p value
Age (y), median (range)	38 (19–74)	45 (20–76)	0.071
Sex (female)	57.0	35.1	< 0.001
Atopy*	43.7	64.3	0.002
Asthma	29.8	44.0	0.019
Current smoker	4.1	5.2	0.681
Nasal congestion/obstruction/blockage	89.3	94.0	0.172
Rhinorrhea/Post nasal drip/Nasal drainage	84.3	69.2	0.005
Purulent nasal drainage	19.8	19.5	0.954
Sinus pressure/pain	73.6	54.9	0.002
Headache/Migraine	25.6	14.3	0.023
Fatigue/Fever/Feel poor	14.9	5.3	0.010
Smell loss/Reduced taste	33.9	72.2	< 0.001
Ear fullness/pain/popping	17.4	13.5	0.399
Eye watering/itching	8.3	6.8	0.651
Cough	24.8	8.3	< 0.001
History of FESS (> 2)	12.4	30.6	0.001
Intra-Operative Pus#	15.0	10.3	0.271
Allergic mucin#	2.7	11.9	0.007

The p value was determined by the Chi-square test.

^*CRSsNP (n=119) and CRSwNP (n=112).

^#CRSsNP (n=113) and CRSwNP (n=126).

CRS Endotyping

We determined mRNA and protein for endotypic markers in ethmoid and NP tissues.⁷ Only a minor subset of patients had matched data for mRNA and protein. Endotyping was blinded to the clinical parameters. For T1 inflammation, we evaluated gene expression and protein levels of IFN-γ. For T2 inflammation, we evaluated CLC mRNA and ECP levels. For T3 inflammation, we measured IL-17A gene expression and protein levels.⁷ If a donor had both mRNA and protein data available, the mRNA results were used. If a CRSwNP donor had both ethmoid and NP tissue available, the NP data was used. We used a cutoff of greater than 90th percentile of the expression in control ethmoid tissue to define endotypes. Control patients were undergoing surgery for non-CRS indicated procedures such as cranial tumor resection or septoplasty, and their clinical characteristics are shown in this article’s online repository (Table E1).

For the analysis of endotype-phenotype associations, we primarily compared clinical presentations with the presence and/or absence of specific endotypes. We secondarily compared untypeable patients, i.e. those who did not have any elevation of T1, T2 or T3 inflammation, with patients who had a single endotype (e.g. they had only T2 inflammatory profile) as well as with patients who had a mixed endotype (e.g. they had both T1 and T2 inflammatory profiles).

CRS Phenotyping

The medical records for each study subject were manually reviewed to obtain information about various clinical and demographic parameters. Reviewers of the medical record were blinded to the endotyping results. Information was gathered from outpatient clinic visits within the 6 months prior to the subject’s sinus surgery and the operative notes from the day of the procedure. Patients were determined to have asthma and/or atopy if there was a physician diagnosis listed in the clinical record. The clinical symptoms reviewed included nasal congestion, obstruction, blockage, rhinorrhea, post nasal drip, nasal drainage, purulent nasal drainage, sinus pressure/pain, headache, fatigue, fever, smell loss, reduced taste, ear fullness, ear pain, ear popping, eye watering, eye itching, and cough. For the data analysis, similar symptoms (e.g. nasal congestion, obstruction, blockage) were grouped together in one category. A history of physician diagnosed migraine was included with the symptom of headache. In the surgical report, two additional parameters, namely the presence of intra-operative pus and allergic mucin, were identified. In some instances, information regarding atopy status or the presence of pus or allergic mucin during surgery was not identified in the medical records. If this information was not available, those patients were excluded from these specific analyses.

Statistical analysis

All statistical calculations were performed using Graphpad Prism version 6.07 (GraphPad Software, La Jolla, CA), SAS version 9.4 (SAS Institute Inc, Cary, NC) and R 3.3.3. The Chi-square test was used for comparisons among different patient groups. The Kruskal-Wallis test with Dunn’s correction was used to compare means among different groups regarding age. Separate logistic regression analyses were performed for the dependent variables intraoperative pus, each specific symptom, history of > 2 FESS, and allergic mucin. The main predictor was endotype with additional covariates including age, sex, asthma, atopy, and NP. The corresponding odds ratio (OR), 95% confidence interval (CI) and p-value were calculated by logistic regression. A p value of less than 0.05 was considered significant.

RESULTS

Inflammatory endotypes in CRS

We analyzed our published data,⁷ which included 121 CRSsNP patients and 134 CRSwNP patients, and defined inflammatory endotypes by using the 90th percentile of expression of markers in control ethmoid tissue as the threshold. Among CRSsNP patients, the overall frequency of having any T1, T2 or T3 inflammation was 21%, 55% and 27%, respectively, of which 1.7%, 34% or 8.3% had evidence of only T1, T2, or T3 inflammation (single endotypes) (Fig. 1). Among CRSwNP patients, the overall frequency of any T1, T2 or T3 inflammation was 17%, 87% and 18%, respectively, of which 0.7%, 62% or 2.2% had only T1, T2, or T3 single inflammation (Fig. 1). Twenty six percent of CRSsNP and 26% of CRSwNP patients had a combination of T1, T2, and/or T3 inflammation and were thus referred to as having mixed endotypes. In contrast, 30% of CRSsNP patients and 9% of CRSwNP patients had no detectable elevations of T1, T2, or T3 inflammation and thus were referred to as having an untypeable (Tun) endotype (Fig. 1). The frequency of T2 inflammation was significantly higher in CRSwNP than CRSsNP (p<0.001) and the frequency of T3 inflammation tended to be higher in CRSsNP than CRSwNP (p=0.073).

Figure 1.

Patterns of inflammatory endotypes in CRS.

Differences in clinical presentation between CRSsNP and CRSwNP

We reviewed 16 clinical parameters, including 14 parameters in the clinical chart and 2 parameters in the surgical report, and examined whether there were any differences between CRSsNP and CRSwNP. We found that CRSsNP had a higher prevalence of females compared to CRSwNP (Table 1). The frequency of asthma and atopy were significantly higher in CRSwNP than CRSsNP (Table 1) confirming our previous findings.^{10, 23} In regards to symptoms, we found that rhinorrhea, sinus pressure/pain, headache/migraine, fatigue/fever and cough were more often reported by patients with CRSsNP compared to CRSwNP patients (Table 1). In contrast, smell loss was more frequently reported by patients with CRSwNP. Additionally, history of recurrent functional endoscopic sinus surgery (FESS) and the presence of allergic mucin were more frequently found in CRSwNP (Table 1).

Inflammatory endotypes associate with clinical presentations in all CRS

We examined whether specific clinical characteristics were associated with a particular inflammatory endotype (Table 2). Among all CRS patients, those with the T2 endotype were significantly more likely to have NPs, asthma, smell loss and allergic mucin and were less likely to report rhinorrhea, cough and pus (Table 2). We also found that the T1 endotype was more common in females, and the presence of pus was significantly associated with the T3 endotype (Table 2). Other features including atopy, nasal congestion, purulent nasal drainage, sinus pressure, headache, fatigue, ear fullness, ocular symptoms, and history of FESS were not different among the 3 inflammatory endotypes (Table E2).

Table 2.

Comparisons of clinical presentations and inflammatory endotypes in all CRS, CRSsNP and CRSwNP

	T1 endotype			T2 endotype			T3 endotype
All CRS	Absence % (n=207)	Presence % (n=48)	p value	Absence % (n=72)	Presence % (n=187)	p value	Absence % (n=198)	Presence % (n=57)	p value
NP	53.6	47.9	0.476	23.6	63.9	< 0.001	55.6	42.1	0.073
Sex (female)	42.0	64.6	0.005	51.4	44.3	0.304	44.9	50.9	0.429
Asthma	36.2	41.7	0.483	26.4	41.5	0.024	38.4	33.3	0.487
Rhinorrhea/Post nasal drip/Nasal drainage	75.7	79.2	0.614	84.7	73.1	0.049	75.6	78.9	0.604
Headache/Migraine	21.4	12.5	0.165	20.8	19.2	0.772	20.8	15.8	0.401
Smell loss/Reduced taste	55.8	45.8	0.211	30.6	63.2	< 0.001	55.3	49.1	0.408
Cough	17.0	12.5	0.446	25.0	12.6	0.016	17.3	12.3	0.368
Intra-Operative Pus#	10.5	20.8	0.053	19.4	9.9	0.046	7.5	30.2	< 0.001
Allergic mucin#	7.3	8.3	0.814	1.5	9.9	0.027	8.1	5.7	0.559
CRSsNP	Absence % (n=96)	Presence % (n=25)	p value	Absence % (n=55)	Presence % (n=66)	p value	Absence % (n=88)	Presence % (n=33)	p value
Sex (female)	53.1	76.0	0.039	60.0	56.1	0.662	55.7	63.6	0.430
Headache/Migraine	28.1	16.0	0.216	14.5	34.8	0.011	27.3	21.2	0.496
Smell loss/Reduced taste	33.3	36.0	0.802	23.6	42.4	0.030	35.2	30.3	0.610
Pus#	11.4	28.0	0.040	21.2	9.8	0.094	8.5	32.3	0.002
CRSwNP	Absence % (n=111)	Presence % (n=23)	p value	Absence % (n=17)	Presence % (n=117)	p value	Absence % (n=110)	Presence % (n=24)	p value
Sex (female)	32.4	52.2	0.072	23.5	37.6	0.258	36.4	33.3	0.779
Headache/Migraine	15.5	8.7	0.400	41.2	10.3	0.001	15.6	8.3	0.357
Smell loss/Reduced taste	75.5	56.5	0.065	52.9	75.0	0.058	71.6	75.0	0.734
Intra-Operative Pus#	9.7	13.0	0.635	13.3	9.9	0.683	6.7	27.3	0.004

The p value was determined by the Chi-square test.

^#All CRS: absence (n=191) and presence (n=48) of T1 endotype, absence (n=67) and presence (n=172) of T2 endotype, absence (n=186) and presence (n=53) of T3 endotype. CRSsNP, absence (n=88) and presence (n=25) of T1 endotype, absence (n=52) and presence (n=61) of T2 endotype, absence (n=82) and presence (n=31) of T3 endotype. CRSwNP, absence (n=103) and presence (n=23) of T1 endotype, absence (n=15) and presence (n=1111) of T2 endotype, absence (n=104) and presence (n=22) of T3 endotype.

We also performed multivariate logistic regression analysis and found that smell loss was associated with T2 endotype (OR 2.80, 95% CI [1.45–5.51], p<0.01) and intra-operative pus was associated with T3 endotype (OR 5.42, 95% CI [2.23–13.61], p<0.01) while controlling for age, sex, NP, atopy and asthma (Fig. 2). In addition, headache/migraine was negatively associated with T1 endotype (OR 0.28, 95% CI [0.08–0.79], p=0.03) (Fig. 2). Other clinical factors were not significantly associated with any endotype when controlling for age, sex, NP, atopy, and asthma (Fig. E1).

Figure 2. Odds ratio for clinical presentations within endotypes in all CRS.

Multivariate logistic regression analysis with predictor endotype controlling for age, sex, asthma, atopy and NP and the corresponding odds ratio (OR), 95% confidence interval (CI) and p-value for headache/migraine (n=230), smell loss (n=230) and intra-operative pus (n=217) in CRS. * p<0.05, ** p<0.01 by logistic regression analysis.

Inflammatory endotypes associate with clinical presentations in CRSsNP

Among patients with CRSsNP, we found that smell loss and headache/migraine were significantly associated with the presence of a T2 endotype (Fig. 3A, Table 2). In addition, the presence of pus was significantly more common in T1 and T3 endotypes (Fig. 3A, Table 2). Other clinical factors were not different among the inflammatory endotypes in CRSsNP (Table E3).

Figure 3. Endotype-phenotype associations in CRSsNP and CRSwNP.

Association of clinical presentations with the absence and/or presence of all T1, T2 or T3 endotypes in CRSsNP (A) and CRSwNP (B). * p<0.05, ** p<0.01 by the Chi-square test.

Inflammatory endotypes associate with clinical presentations in CRSwNP

We next analyzed CRSwNP patients and found that the T3 endotype was significantly associated with the presence of pus and the T2 endotype tended to be associated with smell loss (p=0.058) (Fig. 3B, Table 2). Interestingly, in contrast to CRSsNP, headache/migraine was lower in the presence of T2 endotype in patients with CRSwNP (Fig. 3B, Table 2). Other clinical factors were not different among the inflammatory endotypes in CRSwNP (Table E4).

Comparisons of clinical presentations between single, mixed and untypeable CRS endotypes

Since 26% of CRS patients had more than one endotype (e.g., CRS with T1 and T2 endotypes; called T1/2CRS), we were able to test whether the presence of mixed endotypes may affect the endotype-phenotype associations. The single T1 CRS endotype was rare (1.8%, n=3, Fig. 1) and was not included in the analysis. We compared the clinical presentations between patients with untypeable CRS (TunCRS) to patients with either a single endotype (T2 or T3) or mixed endotypes (T1/T2, T1/T3, or T2/T3) (Table 3). We again found that the presence of NPs, asthma, and smell loss significantly associated with the presence of a single T2 endotype (T2CRS) (Table 3). In contrast, ocular symptoms were negatively associated with T2CRS. The presence of pus and purulent nasal drainage were both significantly associated with a single T3 endotype (T3CRS) (Table 3). Other clinical factors were not different among the single inflammatory endotypes (Table E5).

Table 3.

Comparisons of clinical presentations between single or mixed and untypeable CRS endotypes in all CRS

		Single or multiple endotype
	TunCRS % (n=48)	T2CRS % (n=124)	T3CRS % (n=13)	T1/2CRS % (n=23)	T1/3CRS % (n=8)	T2/3CRS % (n=22)
NP	25.0	66.9	23.1	47.8	12.5	59.1
p value		< 0.001	0.886	0.054	0.438	0.006
Asthma	25.0	42.7	15.4	43.5	50.0	36.4
p value		0.031	0.465	0.115	0.147	0.329
Purulent nasal drainage	18.8	16.3	53.8	21.7	12.5	18.2
p value		0.697	0.011	0.767	0.669	0.955
Smell loss/Reduced taste	31.3	65.9	38.5	52.2	12.5	63.6
p value		< 0.001	0.623	0.089	0.277	0.011
Eye watering/Itching	16.7	6.5	0.0	4.3	12.5	4.5
p value		0.040	0.114	0.114	0.766	0.160
Intra-Operative Pus#	6.8	5.2	50.0	21.7	50.0	26.3
p value		0.687	< 0.001	0.074	0.001	0.033

The p value was determined by the Chi-square test compared to TunCRS. TunCRS: undetectable elevations of T1, T2, or T3 406 inflammation.

^#TunCRS (n=44), T2CRS (n=116), T3CRS (n=12), T1/2CRS (n=23), T1/3CRS (n=8) and T2/3CRS (n=19).

In some instances, mixed endotypes shared similar phenotypes as single endotypes. For example, pus was present in T3CRS as well as T3 mixed endotypes (T1/3CRS and T2/3CRS) (Table 3). Smell loss and NP were present in T2CRS as well as T2 mixed endotypes (T1/2CRS and T2/3CRS), although T1/2CRS (p=0.089 and 0.054, respectively) did not reach significance (Table 3). Similar to T2CRS, asthma tended to be associated with a T2 mixed endotype (T1/2CRS), however, this did not reach significance (Table 3). In contrast, purulent nasal drainage was present in the T3 single endotype but not in T3 mixed endotypes (Table 3). This suggests that the phenotype of mixed endotypes are not necessarily a simple combination of individual endotype- associated phenotypes in CRS.

Finally, we subdivided CRS into CRSsNP and CRSwNP and analyzed the relationships between clinical presentations and either single or mixed inflammatory endotypes in both populations. However, due to low patient numbers, T1 and T3 single endotypes were excluded from the CRSwNP analysis. Among patients with CRSsNP, those with a single T3 endotype (T3sNP) or a mixed T1/T3 endotype (T1/3sNP) were more likely to have pus (Fig. 4A). Among patients with CRSwNP, those with a mixed T2/3 endotype (T2/3wNP) but not with a single T2 endotype (T2wNP) were likely to have pus (Fig. 4B). In addition, smell loss was significantly associated with the single T2 endotype (T2wNP) as well as the mixed T2/T3 endotype (T2/3wNP) (Fig. 4B). A higher prevalence of asthma was found in the mixed T1/T2 endotype (T1/2wNP) with a trend towards being elevated in the single T2 endotype (p=0.066) (Fig. 4B). Interestingly, headache and ocular symptoms were negatively associated with T2wNP and were 279 most common associated with the untypeable CRSwNP endotype (TunwNP) (Fig. 4B).

Figure 4. Endotype-phenotype association in single, mixed and untypeable endotypes of CRSsNP and CRSwNP.

Comparison of clinical presentations between untypeable, single or mixed endotypes in CRSsNP (A) and CRSwNP (B). * p<0.05, ** p<0.01 by the Chi-square test.

DISCUSSION

The division of CRS into two major phenotypes, CRSsNP and CRSwNP, is widely accepted, and the patterns of inflammatory endotypes in these two CRS phenotypes are known to be different.^{7–9, 11} While associations between endotype and phenotype in asthma have been well studied in the past decade,^{16–21, 24} these associations in CRS have only begun to be examined. Tomassen et al. found that the presence of T2 inflammation with Staphylococcus aureus enterotoxin–specific IgE was highly associated with CRSwNP and asthma.⁸ Likewise, Turner et al. found that T2 inflammation was associated with presence of NPs and asthma.²² Although previous studies identified this association between T2 inflammation and presence of NPs and asthma, it was not known whether T2 inflammation was associated with other clinical presentations, or whether T1 and T3 inflammation were associated with other distinct clinical phenotypes. To our knowledge, this is the first study to report how T3 inflammation influences clinical presentations in CRS. Furthermore, we have identified new associations between T2 inflammation and clinical presentations in CRS.

In the present study, we first confirmed the published observations that T2 inflammation was associated with the presence of NPs and asthma comorbidity in CRS (Table 2). This suggests that CRS patients in this study were typical of the general CRS population in the US. Interestingly, we also found that the T1 and T2 mixed endotype (T1/2wNP) showed the highest asthma comorbidity in CRSwNP but not in CRSsNP (Fig. 4B and not shown). Although we did not investigate asthma phenotypes in the present study, CRSwNP with a mixed T1 and T2 endotype may correspond to a unique phenotype of asthma. Further studies will be required to identify the association between asthma and CRS phenotypes and endotypes as well as the endotype of asthma associated with CRS endotypes.

In addition to the presence of NP and asthma comorbidity, we identified that smell loss was strongly associated with the T2 endotype. We recently found that complaints of smell/taste loss were significantly higher in eosinophilic CRS (78% were patients with CRSwNP) and that expression of an eosinophil marker, CLC, in superior turbinate correlated with olfactory defect in CRS.^{25, 26} These studies may suggest that T2 inflammation promotes smell loss in CRS. However, smell loss is also well known to be associated with the presence of NPs,¹ and therefore it was not clear whether the T2 inflammation was just a marker of NPs or whether the T2 inflammation was instead more directly associated with smell loss. In our present study, we found that T2 inflammation was also associated with smell/taste loss in CRSsNP (Fig. 3). Furthermore, smell loss was still associated with T2 endotype while controlling for asthma and presence of NPs (Fig. 2). These results suggest that the symptoms of smell/taste loss may be directly induced by T2 inflammation. It will be valuable to investigate how T2 inflammation mediates olfactory defects.

In contrast to the presence of asthma and olfactory defects, headaches and migraines were differentially reported in CRSsNP and CRSwNP patients with a T2 endotype. We found that the presence of T2 inflammation positively associated with headache and migraine in CRSsNP while they negatively associated in CRSwNP (Fig. 3). This suggests that a subset of endotype-phenotype associations is affected by the presence (or absence) of NPs.

This study also provides the first evidence suggesting that T3 inflammation is associated with the presence of pus and purulent nasal drainage in CRS. The intraoperative finding of pus was almost always associated with T3 inflammation in both CRSsNP and CRSwNP. Multivariate analysis also supported the direct link between T3 inflammation and pus in CRS (Fig. 2). However, documentation of symptoms of purulent nasal drainage was only associated with the T3 single endotype (Table 3). Since endoscopic sinus surgery facilitates the direct inspection of sinus contents, we believe it is a more sensitive measure of purulence than relying on the patient’s ability to identify purulent nasal discharge and may explain why purulent nasal drainage was less strongly associated with T3 inflammation.

CRS is one of the diseases for which antibiotics are commonly prescribed and accounts for 7.1% of all primary diagnoses for ambulatory care visits with antibiotic prescriptions, despite limited evidence of efficacy.²⁷ This level of overuse may lead to the emergence of antibiotic resistant bacteria. Importantly, T3 (Th17) immunity is very important for protection against bacteria and fungi.^{28, 29} In addition, purulent nasal drainage and pus are common symptoms associated with bacterial infections, and we found that these symptoms were associated with the T3 endotype in CRS. Together, our findings may suggest that the T3 endotype in CRS is most strongly associated with bacterial infection, and, importantly, that T3CRS patients may be the most responsive to treatment with antibiotics. This possibility may generate a more precise and personalized medical strategy for this subset of CRS patients. Several studies have suggested a central role for IL-17, Th17 cells and group 3 innate lymphoid cells (ILC3s) in severe neutrophilic asthma that is generally characterized by steroid insensitivity.³⁰ From this we can infer that T3CRS may also be characterized by neutrophilic inflammation in the sinonasal mucosa and be resistant to glucocorticoid treatments. Future study will be required to identify whether the T3 endotype is associated with bacterial infection in CRS, whether T3CRS patients are responsive to antibiotic treatment and whether T3CRS patients have neutrophilic inflammation with resistance to glucocorticoid treatments.

We found that 30% of CRSsNP and 9% of CRSwNP patients were untypeable using the current endotyping markers (Fig. 1). Interestingly, we also found distinct clinical presentations in these untypeable patients, especially in CRSwNP. Ocular tearing and pruritus were more common in TunCRS compared to T2 and T3 single endotypes in all CRS (Table 3) and compared to T2 single and T2 mixed endotypes in CRSwNP (Fig. 4B), although some comparisons did not reach significance. In addition, headache/migraine was significantly associated with TunwNP (Fig. 4B). These results suggest that the untypeable group may have one or more unrecognized endotypes independent of T cell-associated cytokines. Further study will be required to examine whether there are any biomarkers (such as non-T cell associated cytokines or chemokines) that could identify potential new endotypes in untypeable CRS.

Although we showed clear evidence that inflammatory endotypes are associated with distinct clinical presentations in CRS, our present study has certain limitations. We have purposefully selected the method used for identification of endotypes in this study. In the case of published asthma studies, most investigators used periostin, CCL26, eosinophil numbers and FeNO level as biomarkers to identify T2 endotype, since the levels of T cell associated cytokines were not high in tissue.^31–33 In our present study however, we used ECP and CLC, which are very sensitive markers of eosinophils, to determine the T2 endotype in CRS and we were able to use the canonical T cell-associated cytokines IFN-γ and IL-17A for the determination of T1 and T3 inflammation, respectively. It is possible that there are some false negative calls in our data set, i.e. patients whose disease was driven by T1 or T3 cytokines but were nonetheless assigned to the untypeable group. Future study is required to identify more sensitive biomarkers of T1 and T3 endotypes in CRS to diminish the possibility of such false negatives. Another limitation is that the study was performed only in patients undergoing surgery, and we did not control for systemic or topical corticosteroids and antibiotics use perioperatively. These medications could have affected our findings. Future studies in a primary population of CRS patients and studies on the effect of medications would be worthwhile.

In conclusion, we report that a subset of clinical presentations were directly associated with inflammatory endotypes in CRS. Identification of inflammatory endotypes may help in the diagnosis of CRS phenotypic groups and may be useful for the future design of more precise and personal medicine strategies that effectively prevent or treat disease in CRS patients.

Highlights Box.

What is already known about this topic?

Clinically, patients with CRS present with a variety of symptoms including rhinorrhea, nasal congestion, smell loss, and/or facial pressure/pain. Additionally, the sinonasal tissue of CRS patients is characterized by a heterogeneous pattern of inflammation.

What does this article add to our knowledge?

This study investigates the associations between various inflammatory endotypes and clinical phenotypes in CRS and demonstrates that certain inflammatory markers can be linked with particular clinical signs and symptoms.

How does this study impact current management guidelines?

The identified association of certain inflammatory endotypes with specific clinical presentations can be useful in the diagnosis of CRS. Identifying and understanding these associations could lead to improved targeted therapies and more personalized treatment options for patients with CRS.

Abbreviations

CI

Confidence interval

CLC

Charcot-Leyden crystal galectin

CRS

Chronic rhinosinusitis

CRSsNP

CRS without nasal polyps

CRSwNP

CRS with nasal polyps

ECP

Eosinophil cationic protein

FESS

Functional endoscopic sinus surgery

NP

Nasal polyp

OR

Odds ratio

T1

Type 1

T2

Type 2

T3

Type 3

TunCRS

Untypeable CRS

TunsNP

Untypeable CRSsNP

TunwNP

Untypeable CRSwNP