Variation in the tumour necrosis factor gene is not associated with susceptibility to COPD

To the Editors:

In a recent issue of the European Respiratory Journal, Tanaka et al. 1 studied polymorphisms in the tumour necrosis factor (TNF) and lymphotoxin A genes with respect to their effect on lung function of smokers, and failed to find any association with chronic obstructive pulmonary disease (COPD) phenotypes. Tanaka et al. 1 acknowledge that their work is not a true case–control study, but that it would be better described as an investigation of genetic contribution to disease severity. There have been several studies of variation in TNF with respect to susceptibility to COPD, although many of these have used relatively small sample sizes and are therefore underpowered, and so are likely to lead to results that cannot be replicated.

As part of a European Union collaborative project, we have studied polymorphisms within the TNF gene in a large collection of well-characterised Caucasian COPD patients (n = 1,018) and control subjects (n = 911). COPD cases and control subjects were recruited from six European centres, as previously reported 2. The characteristics of each group are summarised as follows, with data expressed as mean±sd, where appropriate. Controls: 63.5% male; age 60.8±8.9 yrs; smoking history 38.6±17.4 pack-yrs; forced expiratory volume in one second (FEV₁) 95.3±10.9% predicted; FEV₁/forced vital capacity (FVC) 77.9±4.9%. COPD cases: 69.6% male; age 65.8±8.2 yrs; smoking history 48.9±23.6 pack-yrs; FEV₁ 43.0±15.3% pred; FEV₁/FVC 47.5±12.2%.

Six single nucleotide polymorphisms (SNPs) in TNF (table 1⇓) were genotyped using Taqman® probes (Geneservice Ltd, Babraham, UK). Primer and probe sequences are available on request. As quality control measures for the genotyping, 2% of samples of known genotype were included and 10% of samples were present in duplicate to check for concordance. All SNPs were in Hardy–Weinberg equilibrium.

The association of single SNPs with COPD was carried out using a Chi-squared analysis; none of the SNPs showed a significant difference in genotype frequency between cases and controls (p≥0.331). Linear regression analysis in cases and controls was used to identify any possible effect of TNF genotype on FEV₁ (with age, smoking history and sex as covariates). Using a stringent cut-off, this also failed to find any significant effect, with the lowest p-value being obtained for rs1800628 in controls only p = 0.010. Allowing for multiple testing, this is unlikely to be a true association.

A case–control analysis of TNF haplotypes was carried out according to 3. Haplotypes of the TNF SNPs are shown in table 2⇓. This analysis identified a total of 10 haplotypes, seven of which were present at a frequency >2%. There was no significant difference in the frequency of these haplotypes between COPD cases and controls (global score statistic = 2.024, 7 degrees of freedom; p = 0.959). Similarly, using the omnibus test performed over all haplotypes in the SAS procedure PROC HAPLOTYPE [4], none of the possible subsets of the six SNPs showed a significant relationship with COPD.

In this large case–control study using well-characterised COPD patients and controls, we failed to find any association of the TNF polymorphisms with the development of COPD. This is in agreement with other smaller studies in Caucasian populations 5–8, although many studies have only considered the -308 SNP. One study, using 169 Caucasian COPD patients and 358 controls, showed an increase of the 489_GA genotype in the COPD patients with an associated odds ratio of 1.9 9; however, this is not replicated in the current study, suggesting that it may have been a false-positive result.

For allele frequencies ranging 0.05–0.20 (using a dominant model and α = 0.01), the current study has >80% power to detect minimum effect sizes of 1.4–1.65. The fact that we see no association with any of the tumour necrosis factor single nucleotide polymorphisms or haplotypes makes it highly unlikely that polymorphisms in this gene play a major role in the susceptibility to chronic obstructive pulmonary disease.