Variations in the TNFα gene and their interactions with the IL4R and IL10 genes in relation to hand osteoarthritis

Osteoarthritis (OA) is the most common joint disorder worldwide and rapidly increasing with ageing populations. OA is a dynamic process involving all the structures within the joint, i.e., cartilage, synovial membrane and subchondral bone. It shows clinical heterogeneity in joint numbers and regions involved. Some patients may have only one site affected (hip, knee, or hand) affected (local OA), while others have clustered joint regions affected in a characteristic distribution (generalized OA) [1].

The hand is among the most frequently affected site in OA [2]. The prevalence of hand OA is higher in women than in men over the age of 50 [3]. Although the pathogenesis of hand OA is largely unknown, familial aggregation and heritability studies indicate a significant genetic role in addition to the involvement of mechanical (repetitive joint loading) and lifestyle related factors (e.g. obesity) [3, 4].

The development and progression of OA are nowadays believed to involve inflammation [5‐8]. Chondrocytes, as well as synovial cells, of OA patients produce increased levels of pro-inflammatory cytokines, which affect metabolism and enhance the catabolism of all joint tissues affected in OA [5]. Among pro-inflammatory cytokines, interleukin-1β (IL1β) and tumor necrosis factor alpha (TNFα) seem prominent and of major importance to cartilage destruction as they are synthesized during the OA process [9‐11]. In vivo studies have shown that these cytokines can act independently or in concert with other cytokines (e.g. IL6) in the induction and propagation of inflammation [12]. Synthesis of the IL1β and TNFα is inhibited by anti-inflammatory cytokines such as IL4, IL10 and IL13 [13]. On the other hand, expression of cytokine genes like IL1 and TNFα is up-regulated in OA [14].

The gene encoding TNFα is located in the class III region of the major histocompatibility complex (MHC) which is the most gene-dense and polymorphic region of the entire genome [15]. TNFα is driving the inflammatory cascade [5]. IL4 in turn is a cytokine produced by T cells, which plays a major role in immunoglobulin E (IgE) production. Its signals are conferred to effector cells through binding to the alpha chain of IL4 receptor (IL4R). IL4 and IL4R are expressed by human articular chondrocytes; data suggest that mechanical stimulation induces the release of IL4 by human chondrocytes after the recognition and transduction of the mechanical signal by integrin [16, 17]. Therefore, the IL4R is an active autocrine or paracrine signaling molecule in a regulatory pathway in the maintenance of human articular cartilage structure and function [16]. Regulation of the structure and function of human articular cartilage occurs by mediating other biochemical responses to mechanical strain, proteoglycan synthesis, or altering the expressions of other extra cellular matrix (ECM) proteins involved in the pathogenesis of OA [16, 17].

So far, the evidence for involvement of genes propagating or inhibiting inflammation in the development or progression of OA is inconsistent, and the observed associations were not replicated in an independent population [18]. Most of the previous studies examined the role of a single gene in OA without taking into consideration the interaction of the genes participating in the regulation of balance between pro- and anti-inflammatory processes. Our group has reported the associations of the IL1 extended haplotype and common IL6 promoter variants with symmetrical DIP OA [19, 20].

The aim of the current study was to investigate the associations of common variants in the TNFα gene and their interactions with variants in the IL4R and IL10 genes in relation to hand OA among middle-aged Finnish women, representing two occupations: dentists and teachers. The possible interactions of the TNFα with variants in the IL1β and IL6 in influencing hand OA were also explored.

The prevalence of hand OA with at least three affected finger joints was 29.5%, being higher among teachers (35.5%) than dentists (24.5%) (Table 1). Participants with hand OA were significantly older and had higher BMI than those without OA.

The genotype frequencies were in HWE in all of the studied polymorphic loci (Table 2). When adjusted for age, occupation and BMI, two TNFα SNPs (“-1031” and “-863”) were associated with hand OA (OR = 1.45, 95% CI 1.01-2.07, p = 0.04 and 1.55, 1.06-2.25, p = 0.02, respectively) (Table 3). Further adjustment for the IL1β and IL6 SNPs had a negligible effect on the observed point estimates, though improving the estimate’s precision (p = 0.03 and p = 0.01, respectively). No statistically significant associations were found between the other two TNFα SNPs and hand OA. Neither were there associations between the SNPs in the IL4R or IL10 and hand OA.

Statistically significant interactions were found between the TNFα “-1031” and IL4R Ser503Pro SNPs, TNFα “-1031” and IL10 “-1082” SNPs, and TNFα “-863” and IL10 “-1082” SNPs and hand OA (p = 0.012, p = 0.0068, and p = 0.02, respectively). The carriage of the TNFα (“-1031”) minor allele was associated with a double risk of hand OA (2.01, 1.26 - 3.22) in women with the IL4R Ser/Ser genotype (Table 4). Similarly, the TNFα “-1031” and “-863” minor alleles were associated with an increased risk of hand OA only in IL10 G/G or A/A homozygotes (2.54, 1.45-4.47 and 2.60, 1.46-4.62, respectively) but not in heterozygotes (G/A).

We also examined the individual and joint effects of the TNFα “-863”, IL1β, and IL6 polymorphisms on hand OA. The risk of hand OA was the highest in the carriers of the minor alleles in all three genes (4.37, 1.84-10.38, p = 0.001). Somewhat lower risks were observed for carriers of the TNFα “-863” minor allele (3.73, 1.28-10.85, p = 0.016) and carriers of minor alleles of IL1β and IL6 SNPs (2.89, 1.29-6.48, p = 0.010).

The degree of pairwise LD between three TNFα SNPs (“-1031”, “-863”, and “-308”) was high; the “-1031” and “-863” SNPs were in complete linkage (D’ = 1, r² = 0.93, p < 0.0001), and the IL4R Ser503Pro and Ser752Ala polymorphisms were also in a strong LD (D’ = 0.998, r² = 0.48, p < 0.0001). The three TNFα promoter polymorphisms composed a total of four haplotypes. The most common of these haplotypes was T-C-G (0.69), followed by G-A-G (0.16), T-C-A (0.13), and G-C-G (0.02). The two IL4R polymorphisms, on the other hand, composed three haplotypes, i.e., Ser-Ser (0.87), Ser-Pro (0.10) and Ala-Pro (0.03).

The TNFα G-A-G haplotype was associated with an increased risk of hand OA when adjusted for age, occupation, and BMI (1.61, 1.10-2.37, p = 0.01) (Table 5). There was no difference between participants with and without hand OA in the IL4R haplotype distribution.

We investigated whether the TNF α promoter polymorphisms are associated with hand OA among Finnish women. The minor alleles of the TNFα “-1031” and “-863” loci, as well as their haplotype, were found to be associated with an increased risk of hand OA. The observed associations were independent of the variants in the IL1β and IL6 genes. Furthermore, our findings suggest that the effect of TNFα polymorphisms on hand OA is modified by the variants within the IL4R and IL10 genes.

A traditional paradigm of OA as a “wear and tear” disease leading to the loss of cartilage has been revised. Nowadays, OA is considered a complex disease with inflammatory mediators released by cartilage, bone and synovium [8]. TNFα is one of the most typical pro-inflammatory cytokines that along with IL1β is connected with cartilage destruction. These two cytokines, which are produced by chondrocytes, mononuclear cells, osteoblasts and synovial tissues, induce the production of a number of inflammatory and catabolic factors [5]. Among the four polymorphic loci studied here, only the TNFα “-308” locus has been shown to affect the TNFα protein levels, the minor allele of the SNP was associated with increased TNFα production in response to various stimuli [27, 28]. However, also opposite observations, e.g., no effect on the protein levels or lowered protein levels, have been reported [29, 30]. When studying the above SNPs with F-SNP-program that is freely available on the internet (http://​compbio.​cs.​queensu.​ca/​F-SNP/​) connected to the main databases, and computationally predicting functional SNPs, all four SNPs are predicted to be functional as they seem to be in the transcription factor binding site [31]. However, the protein level alteration by the studied SNPs still remains unsolved and needs to be further studied.

The few studies that examined associations between TNFα polymorphisms and knee or hip OA have given conflicting results [32]. To our knowledge, this is the first study to report an association between the TNFα variant alleles and hand OA. TNFα can act independently or in concert with other cytokines (e.g., IL1β, IL4, IL6, and Il10) to initiate and expand inflammation [5]. Ignorance of the complex interrelationships between pro- and anti-inflammatory cytokines might be the reason for failure to detect an association between the TNFα polymorphisms and OA. It has been suggested that the combined use of information from multiple markers may be more effective to reveal the association between a genomic region and a trait than a single marker analysis [33].

Previously our group reported the associations of the IL1 and IL6 gene polymorphisms with hand OA [19, 20]. The current findings suggest that TNFα promoter polymorphisms may increase the risk of hand OA independently of the polymorphisms in the IL1 and IL6 genes, and that the effect attributed to combination of variants in all three genes is larger, but less than additive.

When we examined the association of the TNFα polymorphisms with hand OA taking into consideration the variants of other cytokine genes, the IL4R and IL10 polymorphisms appeared to act as effect modifiers.

Vargiolu et al. [34] reported an association of genetic variants in the coding region of the IL4R gene with hand OA among men and women in the age range of 41 to 84 years. However, we failed to replicate this association among our participants. Differences between the study populations, OA phenotypes, and minor allele frequencies might be the reasons for discrepancies in the findings. Our participants were younger (mean age 53 years) than in the study by Vargiolu and coworkers, which may partly explain the difference of the prevalence of hand OA between their and our study (55.6% and 29.5%, respectively).

Naturally occurring anti-inflammatory cytokines such as IL10 inhibit the synthesis of IL1 and TNFα [13]. The IL10 “-1082” polymorphism has been shown to affect the level of the protein production: the A-allele is connected with a significantly higher protein production than the G-allele [35]. This polymorphism was also associated with rheumatoid arthritis in a meta-analysis [36]. As to its role in DIP OA, no association was found in the Dutch population [37]. Similarly, the IL10 “-1082” did not associate with hand OA in our study.

A major strength of the study was that all study participants were of the ethnically relatively homogenous Finnish origin. Each ethnic group has its own set of environmental and genetic factors that contribute to the disease risk, and differences in allelic frequency often affect our ability to detect a susceptibility allele. The Finnish population is known to be a genetic isolate, which originated from a small founder population some 2000 years ago. Therefore, the Finnish population with the relatively homogenous gene pool [38] offers an optimal material for association studies.

Another strength of our study is that we analyzed haplotypes in addition to SNPs. Grouping of SNPs in haplotypes generally leads to a stronger association with the phenotype than individual polymorphisms.

Further, the prevalence of hand OA was similar to that seen in other studies [39‐41], and major potential confounders were controlled for in the statistical analyses.

A limitation of our study is the relatively small number of participants, leading to reduced power to detect small effects and an increased likelihood of spurious findings. This needs to be considered while interpreting the observed associations. Another obvious limitation is the fact that our study participants were all women and consequently the results cannot be generalized to men.

Our results suggest that variants in the TNF α gene play a role in the etiology of hand OA in Finnish women. In addition, the findings are suggestive of a gene-gene interaction of the TNF α with the IL4R and IL10 genes. However, these findings should be considered with caution until replicated in other study population.