TNFA-863 polymorphism is associated with a reduced risk of Chronic Obstructive Pulmonary Disease: A replication study

Background TNF-α mediated inflammation is thought to play a key role in the respiratory and systemic features of Chronic Obstructive Pulmonary Disease. The aim of the present study was to replicate and extend recent findings in Taiwanese and Caucasian populations of associations between COPD susceptibility and variants of the TNFA gene in a Spanish cohort. Methods The 3 reported SNPs were complemented with nine tag single nucleotide polymorphisms (SNP) of the TNFA and LTA genes and genotyped in 724 individuals (202 COPD patients, 90 smokers without COPD and 432 healthy controls). Pulmonary function parameters and serum inflammatory markers were also measured in COPD patients. Results The TNFA rs1800630 (-863C/A) SNP was associated with a lower COPD susceptibility (ORadj = 0.50, 95% CI = 0.33-0.77, p = 0.001). The -863A allele was also associated with less severe forms of the disease (GOLD stages I and II) (ORadj = 0.303, 95%CI = 0.14-0.65, p = 0.014) and with lower scores of the BODE index (< 2) (ORadj = 0.40, 95%CI = 0.17-0.94, p = 0.037). Moreover, the -863A carrier genotype was associated with a better FEV1 percent predicted (p = 0.004) and a lower BODE index (p = 0.003) over a 2 yrs follow-up period. None of the TNFA or LTA gene variants correlated with the serum inflammatory markers in COPD patients (p > 0.05). Conclusions We replicated the previously reported association between the TNFA -863 SNP and COPD. TNFA -863A allele may confer a protective effect to the susceptibility to the disease in the Spanish population.


Background
Tumour necrosis factor alpha (TNF-α) is critical in the regulation of inflammation by inducing a cascade of other inflammatory cytokines, chemokines, and other growth factors [1]. The results of several in vitro and in vivo investigations indicate that high-level generation of TNF-α leads to an exacerbation of the inflammatory and pro-oxidative responses that are important in the pathogenesis of many diseases. TNF-α mediated inflammation is thought to play a key role in both the respiratory and systemic features of Chronic Obstructive Pulmonary Disease (COPD) [2].
Higher concentration of TNF-α has been reported in induced sputum in stable COPD patients and during exacerbations. This cytokine was also found increased in bronchial biopsies and bronchoalveolar lavage fluid of COPD patients when compared with control subjects [3]. A meta-analysis study found higher circulating TNF-α levels in COPD patients when compared to controls [4] and these were also related with important weigh loss in severe patients [5].
Genetic variation in the promoter region of TNFA (chromosome 6p21) has been associated with different phenotype expressions, and a wide range of autoimmune, infectious, and oncologic diseases [2]. A common G-to-A single nucleotide polymorphism (SNP) at position -308 (rs1800629), directly affects gene regulation and has been associated with increased transcriptional activity of TNFA in various disorders [6,7]. This polymorphism has been studied in several COPD related phenotypes, with contradictory results. It has also been associated with chronic bronchitis and an increased risk of airflow obstruction in Asian populations [8,9]. Most studies in Caucasian populations could not demonstrate this association [10][11][12][13]. Only two studies showed an association of the -308 SNP and COPD, the first being a large family study from the Boston early-onset COPD study (2006) and the second performed by Gingo and colleagues (2008) found an increased risk of the A allele of being associated with the disease [14,15].
A 2010 study performed in Taiwanese individuals identified TNF -863A (rs1800630) variant to be associated less frequently with COPD [16].
The neighbouring gene, lymphotoxin alpha (LTA) has also been related to alter TNF-α expression. TNF-α and LT-α are proinflammatory cytokines coded for by the TNFA and LTA genes, respectively, mediating a large variety of inflammatory and immunostimulatory responses. The +252 (A to G) SNP in the LTA gene has received attention due to its supposed implication in gene regulation [17] and reported associations with several diseases [18,19]. Higher levels of TNF-α have been reported to be associated with SNPs in the LTA gene [20]. A case-control study performed on a Russian population reported +252G allele to be associated with a subgroup of severe COPD patients [21]. There is only one study exploring the association of SNPs in both genes and lung function in smokers with mild-to-moderate airway obstruction [12].
The aim of the present study was to replicate recent findings in Taiwanese and Caucasian populations of associations between COPD susceptibility and variants (rs1800629, rs361525 and rs1800630) of the TNFA gene in a Spanish cohort. We further aimed to extend on these findings by assessing whether variants/haplotypes of the entire TNFA and LTA genes may influence COPD susceptibility, severity and lung function decline in a Spanish cohort.

Subjects
Caucasian individuals included in this study were divided into three groups, COPD patients, smokers without COPD and healthy controls. COPD outpatients were recruited from a pulmonary clinic at the Hospital Universitario N. S. de Candelaria (Canary Islands, Spain) from August 2002 to July 2009. Patients with a wide range of disease severity were included using the following criteria: age > 40 years, smoking history > 10 packyears and a post-bronchodilator FEV 1 /FVC ratio < 0.70 measured 20 min after the administration of 400 mg of inhaled albuterol. Pulmonary function tests were measured according to GOLD guidelines [22]. The arterial oxygen tension (PaO 2 ) was measured at rest. Severe hypoxemia was defined by PaO 2 values < 60 mmHg. Exercise capacity was also tested using the best of two 6-min walking distance (6MWD) tests separated by at least 30 min following the ATS recommendations [23]. The body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters. The BODE (Body Mass Index, Airflow Obstruction, Dyspnea, Exercise Performance) Index was calculated as previously described [24]. All patients were clinically stable (no exacerbation for at least 2 months) at the time of evaluation. Patients were excluded if they had history of other diseases like asthma or bronchiectasis. As a control group we included 432 individuals without respiratory disease and no smoking history from an ongoing general adult population cohort named CDC project in the Canary Islands [25]. Finally, the third group included 90 current smokers with a cumulative smoking history of > 15 pack-years and normal lung function (FEV 1 % pred > 0.80; FEV 1 /FVC > 0.70). In order to minimize any possible effect of population structure on our estimations, cases and controls were recruited if they have at least two generations of Canarian ancestry [26].
The Ethics Committee of Clinical Investigation of the Hospital approved this study (approval N°36) and written informed consents were obtained from all subjects.

SNPs selection and genotyping
We selected the haplotype-tagging SNPs in the TNFA and LTA genes using the resequencing data of the European individuals from the SeattleSNPs database http:// pga.gs.washington.edu. A multimarker tagging algorithm with criteria of r2 > 0.8 and minor allele frequency (MAF) ≥ 0.1 were used with the Haploview program (version 3.2) [27]. Eleven SNPs in the TNFA gene (Gen-Bank, AY066019) and LTA gene (GenBank AY070490) were chosen. TNFA -857, -376 and -238 SNPs (rs1799724, rs1800750 and rs361525) were included in the study because they had been reported as susceptibility locus in other diseases [28][29][30].
Venous blood was collected from each individual and total genomic DNA was extracted using the commercially available GFX Kit (GE Healthcare). All tSNPs selected were genotyped using SNaPshot ® SNP Genotyping Assay (Applied Biosystems, Foster City, CA). Genotyping was blind to case or control status of samples. A random 20% of the samples were genotyped in duplicate for quality control. The concordance rate for the duplicate samples was 100%. Moreover, as a confirmational method a random number of samples (patients and control subjects) was genotyped by direct sequencing on an ABI PRISM 310 Genetic Analyser (Applied Biosystems). Samples not yielding the genotypes of all SNPs were excluded from analysis; so 721 individuals remained in the study (199 COPD cases).

Statistical analysis
Demographic and main clinical data are presented as percentage and mean ± SD. Differences among groups were explored using Student t test, Mann-Whitney U rank test, or Pearson χ2 test as appropriate. To perform the analysis we used the statistical package SPSS v17 (SPSS, Inc). Evaluation of the progression of the disease of clinical and pulmonary function variables using the SNPs as comparative factor was performed using a general linear modelling for repeated measures (GLIM). For analysis of severity, GOLD stage and BODE index were dichotomised to a mild-to-moderate (GOLD stage I-II and BODE < 2) and severe disease (GOLD stage III-IV and BODE ≥ 3) (table 1). Allele and genotype distribution in the patients and control groups was compared using contingency tables and χ 2 test for independence. Disequilibrium values, departures from Hardy-Weinberg equilibrium and haplotype reconstruction from biallelic polymorphisms, were calculated using the GENEPOP v3.4, Haploview v3.2 and PHASE v2.1 software programmes [29][30][31][32] as appropriate. A generalized additive model was first used to estimate the genotype relative risks (OR and their 95% confidence intervals) for COPD susceptibility. This was calculated by using SNPSTATS [33]. The effective number of independent marker loci, based on pairwise LD between genotyped SNPs for populations independently, was used for adjustment for multiple tests [34]. To test haplotype associations and for checking for consistency of the inferred data, haplotypes were reconstructed from the best average goodness-of-fit output of six PHASE 2.1 runs with 1,000 permutations [32]. Multivariate logistic regression analysis was used to adjust for potential risk and confounding factors (age, sex, pack-yrs of smoking). ANOVA was performed to test the effect of the polymorphisms and serum proteins levels. In every case, a two-sided p value of < 0.05 was considered statistically significant. Table 1 summarizes the demographic data and baseline characteristics of the study groups. Of the 202 stable COPD patients (153 men and 49 women), 59% of patients were in GOLD stages I-II and 41% in GOLD stages III-IV. Similarly, of the 163 patients where the BODE index was possible to calculate, 68% scored < 2 in the BODE index and 32% scored higher (BODE ≥ 3).

Results
All SNPs conformed to HWE with the exception of rs746868 (p < 0.05), which was then excluded from further analyses. The values for the linkage disequilibrium (LD) between the SNPs in the TNFA and LTA genes are shown in Figure 1. Very strong levels of LD can be observed between SNPs within both genes. SNP -863C/A in the TNFA gene was significantly associated with COPD susceptibility as previously reported [16]. The frequency of the -863 A carrier genotype was significantly lower in COPD patients when compared with healthy controls yielding an odds ratio of 0.61 (95% CI = 0.42-0.88, p = 0.008) under a dominant model of inheritance (see Additional file 1). This result remained significant after adjusting for possible confounding factors as age and gender (OR = 0.50, 95% CI = 0.33-0.77, p = 0.001) in a multivariate analysis (table 2). The -863A allele resulted associated with less severe forms of the disease (GOLD stages I and II) when compared with severe COPD (GOLD stages III and IV) (ORadj = 0.303, 95%CI = 0.14-0.65, p = 0.014) (Figure 2). This was consistent with the finding that the -863A allele was significantly associated with patients who scored lower in the BODE index (BODE < 2 vs. BODE ≥ 3; ORadj = 0.40, 95%CI = 0.17-0.94, p = 0.037) (table 3). Moreover, when analysing longitudinal clinical data of the patients cohort, it resulted that the -863 SNP was significantly associated with FEV 1 percent predicted (FEV 1 % pred) (p = 0.004) and the BODE index (p = 0.003) over the 2 yrs follow-up period. The relation found between these clinical parameters and the -863 genotype did not vary among these groups over time (p > 0.05). In addition we found that FEV 1 % pred and the BODE index differed significantly in each one of three moments included in the longitudinal analysis (table 4). The obtained p-values were corrected for multiple testing by the effective number of independent tests (8.85) resulting in a significant association for allele A carriers of the -863 SNP between the COPD cases and the healthy controls group (adjusted p = 0.02). No significant differences were found in COPD patients versus smokers without the disease when analysing single polymorphisms. Global tests for haplotype were predicted from genotypic data. These analyses revealed no differences in frequency distribution between cases and the controls groups: cases vs. smokers without the disease (p = 0.54) and cases vs. healthy controls (p = 0.35) (table 5). When we calculated haplotype relative effect to the most common haplotype no increased risk/susceptibility was observed for the COPD patients sample. However, the haplotype that contains the -863A allele was less frequently observed in COPD cases than in healthy controls individuals (non significant). In contrast, no association was found with the previously reported [14,15] SNPs rs1800629, rs361525 in TNF gene.

Discussion
Our objective was to replicate results from previous studies [14][15][16] showing association between TNF polymorphisms and COPD. The present study investigates the association of TNFA and LTA genes polymorphisms and COPD presence and severity including eleven SNPs of two genes that cover most of their variability. The most important finding of our research suggests that the previously reported [16]TNFA -863A allele is less frequent in COPD patients. The presence of this allele was associated with a less severe form of the disease (GOLD stages I-II) and with lower scores of the BODE index (< 2). Moreover, this association remained unchanged after adjustment for confounding factors like age and gender, suggesting a robust effect. The most recent study, performed in Taiwanese individuals, found that the -863A allele was associated less frequently with COPD, with an increased FEV1/FVC ratio and higher BMI among heavy smokers [16]. Another two studies on COPD found an association of the haplotype containing the -863A variant with a lower risk for susceptibility to COPD, one performed on a Greek population [35] and the second performed on a Caucasian one [14]. Similarly, in our study the only haplotype that contains the -863A allele also tended to be less likely in COPD patients. This region of the TNFA promoter seems to affect transcription factors binding capacity. The variant -863 has been reported to specifically reduce 10 fold the binding affinity of the transcription nuclear factor (NF-B) specially the form p-50-p50 [36]. Most of the inflammatory proteins that are up regulated in airways of COPD patients are regulated by the transcription nuclear factor NF-B that is activated in alveolar macrophages of COPD patients [37]. The -863A allele has been reported associated with elevated TNF-α production by peripheral blood mononuclear cells stimulated with concavalin A [38] while others have associated this variant with reduced circulating levels of the cytokine [30,39]. In the present analysis we could not find any relation between the SNP and serum levels of TNF-α or any other inflammatory cytokine measured.
One strength of the present study is that the association found for the -863 SNP and FEV 1 and BODE index was also replicated when we explored the SNP association in the three longitudinal clinical measures registered over the two years follow up period. This is the first report of a case-control study that evaluated the influence of gene variants on clinical and pulmonary function longitudinal data in a cohort of COPD patients.
However, we did not replicate other previously reported [14,15] associations of TNF polymorphisms with the disease. Consistent with the majority of previous studies on Caucasians we found that the most studied TNFA polymorphism (-308G/A) was not associated with the presence of COPD in our sample. Only two studies on Caucasians found significant association between this SNP and COPD presence. The first was a family based research from the Boston Early Onset COPD study [15] that analysed 17 families but these findings could not be replicated in a case control study of later onset COPD patients. The second, a case-control study [14], found that the -308A allele had a higher risk of being associated with COPD and was also  associated with worse FEV1/FVC. In that study the authors included as cases a subgroup of 11 individuals with emphysema that did not met COPD diagnosis criteria, and therefore this subgroup may have influence the final results. In addition, a meta-analysis study performed by Smolonska and colleagues [40] reported that TNFA was found associated with COPD susceptibility only in Asians.
In relation to other SNPs within the two genes analysed in previous studies, none of them were found associated with COPD in the present study. This finding is supported by the results obtained by the majority of researches performed in Caucasian individuals [10][11][12][13][14][15]40].
The present study has several limitations. Firstly, the reduced sample size of the smoking control group without the disease cannot exclude that the lack of association found between cases and smokers without COPD with the -863 SNP could be a false negative result due to a type II error. The present analysis should be tested in a large sample of smoking controls. Second, population stratification should be addressed in this kind of studies. Even though, it is known that human populations from Spain, including Canary Islands, are highly homogenous in their genetic background [26], in order to reduce the possibility of subtle population stratification we included 100% Canarian individuals in our control groups (smokers and non-smokers) with at least two generations of ancestors from the islands. And third, we have not find correlation between the genetic variants analysed and the functional aspects of the progression of the disease probably due to the short observational period of our study. Mayor long-term longitudinal studies are needed in order to confirm the relationships of certain factors with clinical and functional variables of the disease.
We have found that COPD patients with the variant -863A presented better lung function (GOLD stages I-II).
The -863A allele may be conferring a protective effect on the progression of the disease, which could be explained by the fact that COPD patients progress in different manner along time. It has been hypothesized that the disease progression could be represented by three steps suggesting that, in most smokers the disease process will not advance if innate inflammation is minimized which situate individuals at step 1 or 2 of the model, which is comparable to smokers with normal lung function or COPD patients at GOLD stage I and II respectively [41]. This may be the consequence of a genetic background that protects smokers in progressing to severe forms of the disease as it is shown by our results.

Conclusion
Our data confirm a previous study where the minor allele of SNP -863 in the TNFA gene was associated with a reduced risk of COPD. Our findings also suggest that the-863A allele is associated with less severe forms of the disease. No association was observed for other SNPs in TNF-LT genes and COPD. This is the first study of TNFA and LTA gene polymorphisms in COPD performed in a Spanish Population.

Additional material
Additional file 1: Table 6. Genotype frequency distribution of the LTA and TNFA genes SNPs in COPD patients, healthy control and smoking control individuals.