These authors contributed equally to this work
Genetic evidence for a role of IL33 in nasal polyposis
Article first published online: 26 OCT 2009
© 2009 John Wiley & Sons A/S
Volume 65, Issue 5, pages 616–622, May 2010
How to Cite
Buysschaert, I. D., Grulois, V., Eloy, P., Jorissen, M., Rombaux, P., Bertrand, B., Collet, S., Bobic, S., Vlaminck, S., Hellings, P. W. and Lambrechts, D. (2010), Genetic evidence for a role of IL33 in nasal polyposis. Allergy, 65: 616–622. doi: 10.1111/j.1398-9995.2009.02227.x
Edited by: Hans-Uwe Simon
- Issue published online: 1 APR 2010
- Article first published online: 26 OCT 2009
- Accepted for publication 14 September 2009
- interleukin 33;
- nasal polyposis;
- single nucleotide polymorphism
To cite this article: Buysschaert ID, Grulois V, Eloy P, Jorissen M, Rombaux P, Bertrand B, Collet S, Bobic S, Vlaminck S, Hellings PW, Lambrechts D. Genetic evidence for a role of IL33 in nasal polyposis. Allergy 2010; 65: 616–622.
Background: Little is known about the genetic factors that contribute to nasal polyposis (NP). A genome-wide association study identified 10 single nucleotide polymorphisms (SNPs) associated with eosinophilia. As eosinophils play a key role in the pathogenesis of NP, we assessed if any of these SNPs contribute to genetic susceptibility of NP.
Methods: We recruited 284 patients with NP in four participating hospitals in Belgium and 427 healthy controls, and genotyped 10 SNPs affecting eosinophilia (rs1420101 in IL1RL1, rs12619285 in IKZF2, rs4431128 in GATA2, rs4143832 in IL5, rs3184504 in SH2B3, rs2416257 in WDR36, rs2269426 in MHC, rs9494145 in MYB, rs748065 in GFRA2, and rs3939286 in IL33) using MALDI-TOF. A two-stage design was used while correcting for multiple testing.
Results: First stage analysis, involving 150 NP patients and 250 controls, identified rs3939286 nearby IL33 as a susceptibility factor for NP. Per at-risk A-allele, rs3939286 increased the risk for NP with an odds ratio (OR) of 1.60 (95% CI = 1.16–2.22; P = 0.0041). Second stage replication analysis in another 123 NP patients and 165 controls confirmed this association (OR = 1.43; CI = 1.00–2.06; P = 0.046). The combined analysis of both stages revealed an OR of 1.53 (CI = 1.21–1.96; P = 0.00041). Given the association of IL33 with NP, we also investigated rs1420101 in IL1RL1, which is the receptor for IL33. Although rs1420101 itself failed to associate with NP, a combined risk assessment of rs3939286 and rs1420101 further increased the risk for NP.
Conclusion: We provide unprecedented genetic evidence suggesting a role for the IL33 pathway in the pathogenesis of NP.
Nasal polyposis (NP) is a chronic inflammatory disease of the nasal mucosa, characterized by the formation of benign grape-like protrusions in the upper nasal cavity leading to nasal congestion or obstruction, rhinorrhea, hyposmia, and facial pain and pressure (1). It is a common disease with a prevalence of up to 4% in the general population (2, 3), for which there is currently no definite cure available. Treatment consists of corticosteroids and functional endoscopic sinus surgery. Recurrence of NP is often seen, with up to 85% of the patients having recurrent polyps within 20 years after surgery (4). The chronic nature of NP and its high degree of recurrence interferes with the quality of life of patients, resulting in a substantial socio-economic burden (5, 6).
Eosinophils are pleiotropic leukocytes involved in the initiation and propagation of inflammatory responses. In NP, eosinophils are the most dominant inflammatory cell type, found in up to 90% of all Caucasian NP (7). Histological studies have shown that eosinophils have a significantly higher activation status in polyp tissue compared with normal nasal mucosa (8). These activated eosinophils are known to release Th2-cytokines (e.g. interleukin-5: IL5), chemokines (such as RANTES and eotaxin-1), and lipid mediators (e.g. leukotriene C4), all of which have pro-inflammatory effects leading to the upregulation of adhesion molecules, activation and regulation of vascular permeability, and mucus secretion (9, 10). Furthermore, by releasing toxic granule proteins, eosinophils can serve as major effector cells inducing tissue damage and dysfunction (9).
An interesting observation is that NP frequently occurs in families. Indeed, some studies reported a 20% prevalence of NP in certain families (11), while others have shown a positive family history in 15% to even 50% of NP patients (12–14). It is not known, however, which factors contribute to this genetic susceptibility for NP.
Thanks to the recent development of genome-wide association (GWA) studies, the identification of genes underlying complex genetic disorders has accelerated (15). Indeed, a large GWA study searching for sequence variants affecting blood eosinophil counts identified five single nucleotide polymorphisms (SNPs) that reached genome-wide significance (IL1RL1, IKZF2, GATA2, IL5, and SH2B3). An additional five SNPs (WDR36, MHC, MYB, GFRA2, and IL33) showed suggestive association with eosinophil count (16).
As eosinophils are pivotal to NP and hereditary factors have been suggested in NP, we here investigated whether these 10 recently identified SNPs predispose to NP.
Materials and methods
We recruited 284 patients of self-declared Caucasian origin with NP in the outpatient clinics of four participating hospitals in Belgium (University Hospital of Mont-Godinne, Yvoir; University Hospital Saint-Luc, Brussels; University Hospitals, Leuven and Sint-Jan General Hospital, Bruges) between August 2007 and March 2009. The diagnosis of NP was based on the EPOS criteria (European Position Paper on Rhinosinusitis and nasal Polyps) (1), being the combination of two or more sinonasal symptoms (nasal obstruction, posterior or anterior rhinorrhea, hypo- or anosmia, headache) in association with bilateral polyps on nasal endoscopy and/or a history of surgery for NP. Exclusion criteria were age under 18 years, presence of cystic fibrosis, known immune deficiency, and unilateral polyp or mass. Concurrent self-reported diagnosis of allergy, asthma, aspirin intolerance, sulfite intolerance, and gastro-esophageal reflux disease was recorded, as well as familial history of NP. Polyps were graded by size and extent in both the left and right nasal fossa on a scale of 0–3, according to the Davos classification (17). Sinus computed tomography (CT) was performed at the discretion of the treating surgeon and was scored using the Lund-MacKay scoring system. The latter relies on a score of 0–2 dependent upon the absence, partial or complete opacification of each sinus system and the ostiomeatal complex, deriving a maximum score of 12 per side (18). The control group consisted of 427 healthy Red Cross blood donors of self-declared Caucasian ancestry recruited from January to March 2008.
Peripheral blood was sampled in both cases and control in K2EDTA tubes and transferred to the Vesalius Research Center (Leuven), where genomic DNA was extracted by a standard salting out method and stored at −80°C until further analysis. Local hospital ethics committees approved this study and all patients and healthy controls provided written informed consent before inclusion and DNA sampling, in accordance with the declaration of Helsinki.
All 10 SNPs were genotyped in a blinded manner using iPLEX technology on a MALDI-TOF based MassARRAY Compact Analyser (Sequenom Inc., CA, USA) as described previously (19). Primers were designed using AssayDesign software (version 184.108.40.206, Sequenom Inc.). The rs4857855 SNP in the GATA2 gene did not fit in the assay design and was therefore replaced by rs4431128, which is a HapMap synonymous SNP that is in complete LD with rs4857855 in the CEU population (r2 = 1). SNP genotypes were assigned with the Typer 4.0 Analyser software (version 220.127.116.11, Sequenom, Inc.). Only samples with ≥80% call success were retained for further analysis. The average SNP call rate was 99.64% with all individual call rates >98.55%. Twenty-six samples were genotyped in duplicate with a 100% concordance.
Data are summarized as frequencies (%) for categorical variables, mean values (± SD) or medians (interquartile range) for continuous variables. Deviation from Hardy–Weinberg equilibrium and case-control association at the allelic level were calculated using PLINK v1.05 (http://pngu.mgh.harvard.edu/purcell/plink/) (20). A statistical threshold of P <0.05 was applied for Hardy–Weinberg disequilibrium testing, whereas a Bonferroni correction was applied for testing 10 different SNPs at the allelic level. Logistic regression was used to calculate the odds ratio of the genotypes for the pooled cohort, while correcting for the individual study group. Heterogeneity across the study populations was calculated using the Chi-square-based Cochran Q statistic. Multinomial logistic regression and Chi-square test were used to calculate the odds ratio and overall P-value of the genotype score. Statistical analyses were performed using spss software version 17.0 (SPSS Inc., Chicago, IL, USA).
Population characteristics and study design
We enrolled 284 NP patients in our study, of which 11 (3.9%) were excluded due to inferior DNA quality (genotype success rate <80%). We also included 427 healthy blood donors from the Red Cross, of which 12 (2.7%) were excluded due to low genotype success rates.
To increase the internal validity of the study, we used a two-stage design to cross-validate and replicate potentially significant associations (21). To this extend, the first 150 NP patients and 250 healthy individuals enrolled in the study were arbitrarily included in the first stage (NP-1), while the remaining 123 patients and 165 controls served as replication in the second stage (NP-2). Baseline characteristics for stage 1 and 2 cohorts, as well as the combined cohort are shown in Table 1. No major differences were seen between both cohorts, and the mean age (56 years), male predominance (69.0%), percentage of patients with asthma (41.4%), aspirin intolerance (18.4%), and family history of NP (18.0%) were in accordance with previous reports (5, 12). As we also included patients who had previous surgery for documented NP, up to 101 (37.0%) patients had a Davos score of 0 on the day of study inclusion. Sinus CT scans were performed at the discretion of the treating surgeon and were accessible in 62 (22.7%) NP patients. As expected, Spearman’s rho correlation coefficients between the sinus CT scan and the endoscopic Davos score were 0.377 (P = 0.003) for the right and 0.450 for the left sinus (P = 0.0002). Overall, this indicates that our NP cohort is in accordance with previously published reports (5, 12).
|Discovery: NP-1 study||Validation: NP-2 study||Total|
|Age, year (IQR)||56 (47–66)||43 (26–51)||55 (42–66)||44 (30–54)||56 (45–66)||43 (27–52)|
|Male, n (%)||101 (67.3)||140 (56.0)||86 (71.1)||86 (53.1)||187 (69.0)||226 (54.9)|
|Asthma, n (%)||60 (40.8)||–||50 (42.0)||–||110 (41.4)||–|
|Allergy, n (%)||47 (32.0)||–||43 (36.1)||–||90 (33.8)||–|
|Aspirin intolerance, n (%)||30 (20.4)||–||19 (16.0)||–||49 (18.4)||–|
|Family history of nasal polyposis, n (%)||22 (15.0)||–||26 (21.8)||–||48 (18.0)||–|
|GORD, n (%)||19 (12.9)||–||15 (12.6)||–||34 (12.8)||–|
|Right, n (%)|
|0||55 (33.3)||–||56 (48.3)||–||111 (43.4)||–|
|1||30 (21.4)||17 (14.7)||47 (18.4)|
|2||33 (23.6)||31 (26.7)||64 (25.0)|
|3||22 (15.7)||12 (10.3)||34 (13.3)|
|Left, n (%)|
|0||52 (37.1)||–||59 (50.9)||–||111 (43.4)||–|
|1||37 (26.4)||23 (19.8)||60 (23.4)|
|2||31 (22.1)||20 (17.2)||51 (19.9)|
|3||20 (14.3)||14 (12.1)||34 (13.3)|
|Sinus CT scan|
|Right, median (IQR)||10 (6–12)||–||8 (6–10)||–||9 (6–11)||–|
|Left, median (IQR)||9 (8–11)||7 (5–10)||8 (6–11)|
The rs3939286 SNP is associated with nasal polyposis
In the first stage (NP-1), we genotyped all 10 SNPs recently reported to be associated with blood eosinophil counts (16). For none of these, the genotype distribution deviated from Hardy–Weinberg equilibrium. Furthermore, all allele frequencies were in accordance with those observed in the HapMap CEU population. When comparing allele frequency of all 10 SNPs between NP patients and controls, the rs3939286 variant was significantly associated with NP (OR = 1.60; 95% CI = 1.16–2.22; P = 0.0041, Table 2). This association remained significant after correction for multiple testing at a preset threshold of P < 0.005. None of the other SNPs was associated with NP. Association of the rs3939286 variant with NP was confirmed in the stage 2 cohort (NP-2). At the allelic level, the risk associated with the A-allele was replicated and exhibited an OR = 1.43 (CI = 1.00–2.06; P = 0.046).
|SNP||Chromosome||Position||Gene||Risk allele||Risk allele frequency in cases (%)||Risk allele frequency in controls (%)||OR (95% CI)||P - value|
|rs1420101||2q12||102 324 148||IL1RL1||A||129 (43.0)||199 (39.8)||1.14 (0.85–1.52)||0.386|
|rs12619285||2q13||213 532 290||IKZF2||G||88 (29.3)||152 (30.4)||0.95 (0.69–1.30)||0.745|
|rs4431128*||3q21||129 734 368||GATA2||T||45 (15.0)||88 (17.6)||0.83 (0.56–1.22)||0.330|
|rs4143832||5q31||131 890 876||IL5||C||240 (80.0)||411 (82.2)||0.87 (0.60–1.25)||0.439|
|rs3184504||12q13||110 368 991||SH2B3||T||154 (51.3)||265 (53.0)||0.94 (0.70–1.25)||0.666|
|rs2416257||5q22||110 463 389||WDR36||G||260 (86.7)||414 (82.8)||1.32 (0.88–1.98)||0.171|
|rs2269426||6p21||32 184 245||MHC||T||104 (34.7)||183 (36.6)||0.92 (0.68–1.25)||0.588|
|rs9494145||6q23||135 474 245||MYB||T||238 (79.3)||377 (75.4)||1.21 (0.85–1.72)||0.276|
|rs748065||8p21||21 734 049||GFRA2||A||203 (67.7)||363 (72.6)||0.79 (0.58–1.08)||0.132|
|rs3939286||9p24||6 200 099||IL33||A||92 (30.7)||108 (21.6)||1.60 (1.16–2.22)||0.0041|
The combined analysis of the NP-1 discovery and NP-2 validation cohorts further confirmed that the risk GA and AA genotypes were significantly increased in the nasal polyp cases compared with controls, with the A-allele conferring a risk for NP with an OR of 1.53 (CI: 1.21–1.96; P = 0.00041, Table 3). There was no significant heterogeneity between the associations in both cohorts (Cochran Q statistic for heterogeneity P = 0.49). At the genotypic level, the rs3939286 SNP followed an additive risk effect. Indeed, compared with homozygous wild type carriers (GG), heterozygous (GA) and homozygous (AA) carriers of the A-risk allele had an increased risk of 1.55 (CI = 1.12–2.14; P = 0.008) and 2.40 (CI = 1.28–4.49; P = 0.006), respectively (overall P = 0.003).
|rs3939286 Genotype||Risk allele frequency (%)||OR (95% CI)||P-value|
|AA (%)||AG (%)||GG (%)|
|NP-1 discovery study||Cases (n = 150)||13 (8.7)||66 (44.0)||71 (47.3)||92 (30.7)||1.60 (1.16–2.22)||0.0041|
|Controls (n = 250)||12 (4.8)||84 (33.6)||154 (61.6)||108 (21.6)||Reference|
|NP-2 validation study||Cases (n = 123)||12 (9.8)||56 (45.5)||55 (44.7)||80 (32.5)||1.43 (1.00–2.06)||0.046|
|Controls (n = 165)||8 (4.8)||67 (40.6)||90 (54.5)||83 (25.1)||Reference|
|Pooled||Cases (n = 273)||25 (9.2)||122 (44.7)||126 (46.2)||172 (31.5)||1.53 (1.21–1.96)||0.00041|
|Controls (n = 415)||20 (4.8)||151 (36.4)||244 (58.8)||191 (23.0)||Reference|
As eosinophils also play an important role in asthma and allergy, and asthma and allergy were present in 41.4% and 33,8% of the NP patients, respectively, we performed a subgroup analysis in NP patients without concurrent asthma or allergy. After stratification for asthma, we found a stronger association of rs3939286 with NP in patients without asthma (overall P = 3.0 × 10−5), with increased ORs for the GA and AA carriers of 2.17 (CI = 1.46–3.22; P = 1.2 × 10−4) and 3.49 (CI = 1.72–7.08; P = 0.001), respectively. No significant association was seen in NP patients with asthma (OR = 1.06; CI = 0.68–1.65; P = 0.81 for GA and OR = 1.37; CI = 0.56–3.40; P = 0.49 for AA carriers; overall P = 0.78). Similar findings were found in NP patients with allergy (overall P = 4.2 × 10−4) with ORs for GA and AA carriers of 1.93 (CI = 1.33–2.79; P = 0.001) and 2.67 (CI = 1.32–5.42; P = 0.006). No significant association was seen in NP patients with allergy (OR = 1.12; CI = 0.69–1.83; P = 0.63 for GA and OR = 1.99; CI = 0.83–4.79; P = 0.12 for AA carriers; overall P = 0.31). These data indicate that the association with NP was not due to a confounding association with asthma or allergy. However, data in the individual subgroups should be interpreted cautiously, due to the reduced number of NP patients present in the subgroups with asthma or allergy (see Table 1).
SNPs in the IL33-IL1RL1 pathway are associated with nasal polyposis
The rs3939286 is located on chromosome 9p24 near the IL33 gene. The IL33 gene product belongs to the IL1-superfamily and is the natural ligand for the interleukin-1 receptor-like 1 receptor (IL1RL1). Interestingly, the rs1420101 SNP, located in the IL1RL1 gene, was also associated with blood eosinophil count (16) and therefore also genotyped in our NP study. Although rs1420101 itself failed to associate significantly with NP (Table 2), it is still possible that this variant together with rs3939286 further increase the risk for NP. To assess this possibility, the number of risk alleles in the IL33 and IL1RL1 SNPs per individual was counted and a genotype score was constructed (ranging from 0 to 4), as previously described (22). The proportion of cases and controls carrying different number of risk alleles for the two loci in the combined cohort is shown in Fig. 1A. Using a logistic regression, we found a significant shift towards a higher number of risk alleles carried by cases compared with controls (P = 2.7 × 10−4), with an OR of 1.38 (CI = 1.16–1.63) per additional risk allele. Compared with carriers with no risk allele, the odds ratios per number of risk allele increased gradually (P = 0.002; Fig. 1B). Overall, this indicates that rs3939286, and to a lesser extent also rs1420101, increase the susceptibility to NP.
In this multicenter study, we investigated the association between NP and 10 common genetic variants, recently shown to affect blood eosinophil counts (16). Using a two-stage design, we found that rs3939286 in IL33 was significantly associated with NP and conferred an increased risk of 53% per A-risk allele. In addition, when generating a genotype score consisting of the total number of risk alleles of rs3939826 and rs1420101 – a SNP in the IL33 receptor or IL1RL1 gene – we found an increased risk for NP of 37% per increase in one of any of the two risk alleles. To the best of our knowledge, this suggests a hitherto unrecognized genetic role of both SNPs in NP and mandates careful follow-up analysis of the IL33-IL1RL1 genes and signaling pathways in NP.
These observations raise a number of questions. For instance, (a) How do these SNPs affect the function of the genes they are located in? (b) What is the biological function of these genes? (c) How do they potentially contribute to NP?
The rs3939286 variant, which was most significantly associated with NP, is located on chromosome 9p24 at a 32 kb proximal distance from the start codon of the IL33 gene. Although this may seem rather distant from IL33, this cytokine is the closest gene to rs3939286 and is therefore the most likely to be affected by rs3939286. The rs1420101 variant is located in the receptor of IL33, i.e. IL1RL1 (also known as ST2) on chromosome 2q12. The rs1420101 variant is a synonymous coding SNP located in exon 6. It is not immediately clear how both SNPs could affect expression or function of IL33 or IL1RL1. However, both SNPs are tagging SNPs that serve as proxies for other SNPs in high linkage disequilibrium, which could affect the function or the expression of both genes. Follow-up studies are therefore necessary to identify the causal SNPs, and confirm that they affect the IL33 and IL1RL1 function and thereby contribute to NP.
IL33 represents a recently described cytokine with several immune functions related to the pathophysiology of NP disease. It is a new member of the IL1-superfamily, which also includes IL1β and IL18. Like IL1β and IL18, IL33 was found to have strong immunomodulatory functions. While IL1β and IL18 promote Th1 responses, IL33 rather induces the production of allergy-mediating Th2 cytokines, such as IL4, IL5, and IL13 (23). The latter cytokines are also known to play a prominent role in the pathogenesis of NP (5).
Of additional interest is the observation that IL33 and ILRL1 are emerging key players in the biology of eosinophils. Indeed, eosinophils have been shown to express the receptor IL1RL1 and to become strongly activated by IL33 (24, 25). In addition, IL33 increases survival of eosinophils and triggers superoxide production and degranulation of eosinophils at least as effectively as IL5 (24, 25), which is one of the most potent eosinophil-selective cytokines. In vivo experiments further demonstrated that mice injected with IL33 exhibit increased blood eosinophil counts and profound pathological changes in the lung and intestine mucosal tissues characterized by eosinophil infiltrates. Besides eosinophils, IL33 also promotes adhesion, cytokine production, degranulation, and survival of mast cells (26, 27). Different cell types have been shown to constitutively express IL33, including human smooth muscle cells, as well as epithelial cells forming the bronchus of small airways. Furthermore, IL33 gene expression can be induced by the pro-inflammatory cytokines TNF-α and IL-1β in primary lung or dermal fibroblasts and keratinocytes (23). Although expression of IL33 has not been investigated in nasal mucosa per se, it is also likely to be expressed in nasal mucosa as the nasal cavity, bronchus, and distal parts of the respiratory conducting system share the same epithelium. Overall, the location of both SNPs in IL33 and IL1RL1 as well as the recently described biological function of IL33 and IL1RL1 suggest that genetic variability in the IL33 signaling pathway contributes to the susceptibility of NP.
Given the importance of eosinophils in the pathogenesis of NP and the 10 SNPs discovered to influence eosinophil counts (16), it might be surprising to see that only one out of 10 SNPs showed significant association with NP. However, although eosinophils are central in many diseases including asthma and NP, these different diseases could rely on distinct signaling cues. It could be hypothesized, for instance, that IL5 would play a more prominent role in asthma, whereas IL33 could be more engaged in NP. This would explain why IL5 inhibition is more successful in the treatment of asthma than NP (28–30). Further studies are certainly needed to determine if this hypothesis is valid. The identification of rs3939286 in IL33 as a strong genetic susceptibility factor for NP is, however, encouraging and could provide incentive to discover additional genetic loci in NP. Based on the successful identification of many novel genes contributing to complex genetic disorders in GWA studies (31), a large GWA scan in NP should be considered as the most straightforward approach.
The strong points of this study are its multicentered design, the diagnosis of NP made by nasal endoscopy and not by a clinical questionnaire, as well as the two-stage design of the study, which allowed internal replication. Furthermore, to the best of our knowledge, this study recruited the largest number of NP patients so far. However, some limitations should be acknowledged. Although our controls were healthy blood donors, NP was not excluded by endoscopy in these individuals. Nevertheless, other groups have used similar approaches (32, 33). Concurrent asthma, allergy, aspirin intolerance, familial history, etc. of the NP patients was obtained through medical history and was not confirmed by state-of-the-art techniques such as spirometry and allergy testing. In addition, although the two-stage design of analysis serves as internal replication, external replication in an independent cohort is needed. It is also not known yet whether these genetic variants are associated with altered IL33 and IL1RL1 protein levels. This study therefore provides a genetic pointer, more than a definite proof, for the involvement of the IL33 pathway in NP. Finally, we did not have peripheral white blood cell counts for the patients and were thereby unable to analyze blood eosinophil count as an intermediate phenotype for the genotyped SNPs.
In conclusion, we found that two SNPs in the IL33-IL1RL1 pathway increased susceptibility to NP. These findings provide hitherto unprecedented genetic evidence for a possible role of IL33 in the pathogenesis of NP.
The authors thank N. Lays & G. Peuteman for their excellent technical assistance and also thank all doctors, nurses, research assistants, patients, and healthy blood donors who participated to this study.