A number of observations indicate that genetic variation in the TLR7TLR8
gene region influences the risk for and the degree of AR: 1) The a priori
genes gave the strongest signals among the tested TLR genes. 2) Association tests in both the Swedish and the Chinese populations yielded several uncorrected p-values <0.05 with their corresponding q-values <0.1 and in some cases <0.05. When haplotype associations were tested using the Bonferroni correction, significant differences were still observed. 3) Comparing the results from the Swedish and Danish populations, a similar pattern of association was seen. In the Danish population [12
], two SNPs were associated with AR, rs179008 in TLR7
and rs2407992 in TLR8
. The A-allele of the rs179008 SNP was over-transmitted in Danish cases (P
= 0.0039). This pattern was seen also in the Swedish population, albeit at a non-significant level (P
= 0.38). In the same way, the G-allele of the rs2407992 SNP was over-transmitted in Danish cases (P
= 0.037) with the same pattern observed in Swedish cases, but again at a non-significant level (P
= 0.40). Thus, the congruence of the allelic associations detected in the Swedish and Danish populations may simply reflect the fact that they have the same major allele. 4) When the effect of genotype on the severity of phenotype among the cases were tested, several p-values yielded uncorrected p-values <0.05 and corresponding q-values < 0.1 and in some cases < 0.05. This is particularly important since the phenotype test relies on cases only and furthermore do not depend on the exact genotype numbers among the cases. Thus, the phenotype test is independent from the case–control association test. In addition, a recent genome wide association study for AR based on genotyping using the Illumina HumanHap 550 k BeadChip observed no SNPs with genome wide significance, and only a few SNPs showing suggestive association in MRPL4, BCAP, CSF1R
]. In concordance with the results of the present study, none of the TLR-genes of the autosomes showed suggestive associations. Since the GWAS did not present data on SNPs from the X and Y chromosomes, no data are available for TLR7 and TLR8 and the results of the two studies cannot be directly compared for these two genes.
Although a number of results indicate an association between genetic variation in TLR7
and occurrence and degree of AR, two discordant observations have been made. The first is that, for many of the SNPs in the region, the major allele in the Swedish population is the minor allele in the Chinese population and vice verse. However, in both populations it is the most common haplotype that is associated with AR. A situation where different major alleles of the same locus are risk-associated in different populations is compatible with a scenario where the risk is contributed by many different rare alleles that each has a relatively limited life span. Under these conditions, the major haplotypes will on average be more associated with risk. There are good reasons to believe that multiple rare variants, both within and across genes, collectively affect the expression and function of genes. This hypothesis can be tested by simply re-sequencing large numbers of cases and controls and evaluate the patterns of rare alleles in the two groups. Many reports on the association of rare variants with specific disease phenotypes exist (reviewed in Bansal et al.
The second discordant observation is that, in the Swedish population the AR association is seen primarily in females, whereas in the Chinese population the association is observed among males only. However, the Swedish population is most likely underpowered when analyzing X-chromosomes in males and a contribution to disease from males can therefore not be ruled out. In fact, when analyzing both males and females together, the P
-values of the significant SNPs (rs3788935, rs3761624, rs4830805) are almost the same compared to when analyzing females separately. Also, when analyzing males, a tendency for association among these SNPs (P
-values from 0.25 to 0.49) can be observed. In the Chinese population, however, the observed association is restricted to males with no tendency of association among females. Also in the study by Møller-Larsen et al.
] several strong sex-specific differences were identified.
A number of studies have reported associations between various diseases and variants in the TLR7
genes. For example, systemic lupus erythematosus was found to be associated with variation in TLR7
] and Crohn’s disease and ulcerative colitis with variation in TLR8
]. In particular, the only high frequency missense polymorphism in the TLR8
gene (rs3764880) has been implicated in a number of diseases, such as tuberculosis [30
] and progression of HIV infection [31
]. This SNP is in perfect LD with rs3761624, indicating that these SNPs may serve as proxies for each other. The common allele of rs3761624 (A-allele in the CEU population) is present on the risk haplotype detected in the Swedish population, being compatible with the reported disease associations for rs3764880 where the common allele is the disease-associated allele. The pleiotropic effect seen for this allele also supports the disease association in the present study. Just as in the case of AR in the present study, several of the studies cited above reveal sex-specific disease associations for the TLR7-TLR8
gene region. In the association study of Crohn’s disease and ulcerative colitis one risk- and one protective haplotype was detected among females, and the study of systemic lupus erythematosus detected an association to this region with a stronger effect in males compared to females (OR, male/female = 2.3, 95 % CI = 1.64-3.30).
TLR7 is expressed by inflammatory and structural cells in both the upper and lower airways. It has a well-defined protective role during viral infection. We have previously demonstrated that TLR7- mediated activation of eosinophils is related to the atopic status of the patient and that the presence of a Th2-like cytokine milieu affects the outcome of the response [32
]. Thus, eosinophil activation via TLR7 might engender a link between viral infection and allergic exacerbations. Others have, using various animal models, demonstrated beneficial effects of TLR7 agonists in allergic asthma [33
] and a recent clinical trial has shown encouraging results, suggesting that TLR7 agonists might be a novel alternative for treatment of allergic rhinitis [34
Association analysis based on case–control populations is a very popular method to search for genes that influence specific phenotypes. However, many studies suffer from limited population sizes, where the obvious drawback is the risk of false positives. This emphasizes the importance of large population sizes and the replication of positive findings but also of meta-studies where the combined efforts of many studies are evaluated. In the present study, genetic variation in the TLR7-TLR8 gene region was associated with AR in one Swedish and one Chinese population. Since this region has earlier been associated with asthma and AR in a Danish linkage study and since TLR7 agonists have shown beneficial effects both in allergic asthma and in allergic rhinitis, this speaks strongly in favour of this region being truly involved in the development of AR.