|Home | About | Journals | Submit | Contact Us | Français|
Endotoxin exposure induces airway inflammation, hyper‐responsiveness and higher expression of tumour necrosis factor (TNF). This study was conducted to investigate whether TNF polymorphisms modify the effect of endotoxin exposure on chronic declines in lung function.
Associations between TNF and LTA polymorphisms, endotoxin exposure and lung function were analysed in 263 cotton workers and 230 silk workers as a reference group, who were prospectively followed for 20 years. Multiple linear regression models were used to assess the association, with adjustment for smoking and other covariates.
Endotoxin exposure was associated with faster lung function decline among genotypes associated with higher TNF expression levels, with estimates of annual FEV1 change in relation to endotoxin exposure of –2.9 ml and –6.8 ml in the G/G and G/A+AA genotypes, respectively, for the TNF polymorphism; and –2.0 ml, –4.0 ml and –3.6 ml in A/A, A/G and G/G genotypes, respectively, for the LTA polymorphism. When joint effects of endotoxin exposure and smoking were considered, the effect modification of TNF and LTA polymorphisms was prominent in never smokers.
TNF and LTA polymorphisms may modify the association between occupational endotoxin exposure and longitudinal lung function decline, which was more clearly observed in never smokers.
Long‐term occupational exposure to vegetable dust may cause lung function decline and airway disease. Both animal and epidemiological studies suggest that bacterial endotoxin is the major causative agent in organic dust that contributes to acute and chronic airway inflammation and airflow obstruction.1,2 The incorporation of endotoxin into macrophages and endothelial cells results in local production of inflammatory cytokines, along with subsequent migration of inflammatory cells into the lung and the penetration of cytokines into the blood.3
Tumour necrosis factor (TNF), an important regulatory cytokine with strong proinflammatory and immunomodulatory properties, elicits the widest spectrum of biological activities, including defense against bacterial infection.4 TNF helps mediate the septic shock state induced by bacterial endotoxin and the wasting diathesis that typifies chronic diseases.5 Natural induction of TNF may have protective effects, but its overproduction may be detrimental and even lethal to the host.4 TNF may increase airway responsiveness and neutrophil infiltration,6 promote tracheal smooth muscle proliferation and alter smooth muscle function,7 suggesting that high TNF expression levels may contribute to the airway inflammation and hyper‐responsiveness.
The polymorphic genes of TNF and LTA (lymphotoxin alpha or TNF beta) are arranged in tandem. The A allele of the TNF –308G/A polymorphism (rs1800629) and the G allele of the LTA 252A/G polymorphism (rs909253) are associated with higher production and expression of TNF8,9 or LTA,10 respectively, and are the most widely studied polymorphisms for these two genes. Some studies found no associations between TNF or LTA polymorphisms and lung function decline or the development of chronic obstructive pulmonary disease (COPD),11,12,13,14,15,16 while several others suggested that the two TNF polymorphisms are associated with the development or prognosis of COPD.17,18,19
In a previous study, we observed that long‐term occupational exposure to cotton dust resulted in faster declines in lung function, which were associated most closely with high concentrations of endotoxin.20,21 It has also been reported that endotoxin exposure may induce increased concentrations of TNF in serum22 and bronchoalveolar lavage.23 Polymorphisms relevant to different TNF serum levels may affect the association between endotoxin exposure and lung function changes. Based on this evidence, we hypothesised that polymorphisms associated with higher TNF expressions increase the effect of endotoxin exposure on lung function changes. In addition, cigarette smoking, a major risk factor for chronic airway obstruction may, in turn, modify the associations among TNF polymorphisms, endotoxin exposure and pulmonary function. The primary aim of this study was to test this hypotheses in a cohort of textile workers, who were followed for 20 years.
The Institutional Review Boards of the Harvard School of Public Health, the Putuo District People's Hospital, and the Human Resources Administration, China approved the study. The study subjects came from a longitudinal cohort study that was undertaken in cotton textile workers in Shanghai, China, whose respiratory health and pulmonary function were followed prospectively for 20 years. Details of this population are described elsewhere.20,24 In brief, 447 cotton textile workers and 472 silk textile workers were recruited and studied from 1981 onwards. Follow‐up surveys were conducted in 1986, 1992, 1996 and 2001. Forced expiratory manoeuvre data were available through 2001. Since cotton dust and endotoxin levels in the silk mill were nearly zero over the period of study, we assigned cotton workers as an endotoxin exposure group and silk workers as a reference group. A total of 501 subjects who returned to the last survey and donated blood samples were included in this analysis, which consisted of 267 (60%) in the exposure group and 234 (50%) in the reference.
DNA was extracted from peripheral blood samples using the Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN, USA). The TNF ‐308G/A and LTA 252A/G polymorphisms were genotyped by the 5′ nuclease assay using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA). The primers, probes and reaction conditions are available upon request. Genotyping was performed by laboratory personnel blinded to case‐control status, and a random 5% of the samples were repeated to validate genotyping procedures.
Differences in the distribution of the principal covariates between the exposure group and reference group were tested using the χ2, Fisher's exact, and Student's t tests, where appropriate. Annual decline in FEV1 was calculated based on the differences in pre‐shift FEV1 measurements between the last and baseline survey and divided by 20 (years). The reason that we used the two‐point FEV1 was because the data at these two surveys were available for all of the study subjects. This enabled us to maintain statistical power to detect possible exposure–response relations, given that there were missing data from 14% to 26% at other follow‐up surveys. Multiple linear regression models were used to determine the association between TNF and LTA genotypes, alone or in combination with endotoxin exposure, and the annual decline rate of FEV1 (treated as a continuous variable). Covariates considered in the models included gender, smoking status (categorical variables), age, height, baseline FEV1 and smoking amount (continuous variables). Interactions between these variables were also examined.
Detection of linkage disequilibrium between the two polymorphisms was based on Lewontin's D' in all subjects.25 Haplotypes of the two ERCC1 polymorphisms were generated using the Partition Ligation‐Expectation Maximization (PL‐EM) version 1.0.26 This software uses an efficient variant of the EM algorithm to reconstruct individual probabilities for individual phasing accuracy, based on unphased genotype data. At the same time, it also provides estimates on the overall haplotype frequencies as well as their standard errors.
In the genotype analyses, the wild G/G genotype of TNF and A/A genotype of LTA polymorphisms were treated as reference groups, and the TNF G/A and A/A genotypes were combined because of sparsity of A/A genotype. TNF and LTA were analysed separately as well as in combination. In addition to determining the overall association between TNF genotypes and longitudinal changes in lung function, we assessed possible effect modification of TNF polymorphisms on the association between endotoxin exposure and lung function by stratifying TNF and LTA polymorphisms in the analysis. Finally, we investigated the joint effects of combined TNF and LTA polymorphisms, endotoxin exposure, and smoking on the longitudinal changes in FEV1, of which non‐exposed, non‐smokers with TNF “wild type” genotype were defined as the reference group. Statistical analyses were carried out using SAS personal computer software (version 9.2, SAS Institute, Cary, NC, USA).
There was no significant difference in age, gender, height, smoking status and baseline FEV1 between subjects with and without blood sample in either group (p>0.05). A total of 493 out of the 501 subjects were genotyped successfully for both TNF (‐308 G/A) and LTA (252A/G) polymorphisms, including 263 in the exposed group and 230 in the reference. The characteristics of the exposed and reference groups are shown in table 11.. No significant differences were found in the demographic features such as age, gender, height, smoking status, smoking amount and baseline FEV1 between the two groups (p>0.05).
Both TNF and LTA polymorphisms were in Hardy–Weinberg equilibrium (p>0.05 by χ2 test). Genotype frequencies in the exposed and reference groups were similar (table 11).). The two polymorphisms were in high linkage disequilibrium (D'=0.98), and subjects with wild types of LTA (A/A genotype) were also carriers of wild types of TNF (G/G genotype). Applying the PL‐EM algorithm, the posterior probabilities of individual haplotypes were all greater than 99%; therefore, we assigned each individual haplotype with the highest posterior probability. Total of 3 haplotypes (haplotype 1, TNF_G‐LTA_A; haplotype 2, TNF_G‐LTA_G; and haplotype 3, TNF_A‐LTA_G) were generated using PL‐EM, with the similar haplotype distributions in the exposed and reference groups (table 11).). Because of the high linkage disequilibrium between the two polymorphisms, all subjects with the LTA A/A genotype had two copies of haplotype 1, subjects with the LTA A/G genotype had one copy of haplotype 1, and subjects with the LTA G/G genotype had no copies of the haplotype 1. Thus, analysis of gene–gene joint effects was performed subsequently, as the haplotype analysis did not provide additional information.
Overall, the average annual changes in FEV1 were –31.5 ml in exposed subjects and –27.5 ml in reference (“–” represents the direction of decline). Regression analysis suggested that endotoxin exposure was associated with –3.2 ml/year change. There was no significant association between TNF and LTA polymorphisms and lung function decline. For the TNF polymorphism, the estimated annual change in FEV1 was –0.3 ml for GA+A/A, relatively to the G/G genotype (p=0.86). For the LTA polymorphism, the annual changes were –0.4 ml (p=0.86) and 0.3 ml (p=0.17), respectively, for the A/G and G/G genotypes when compared with the A/A genotype.
To determine potential effect modification of TNF or LTA polymorphisms on the association between endotoxin exposure and annual changes in FEV1, we fitted multiple liner regression models with TNF, LTA polymorphism, as well as their combination, respectively. Age, sex, height, baseline FEV1, smoking status and smoking amount were considered in the initial models as candidate covariates or confounders. Height and sex were removed from the final models because of little contribution to the outcomes of interest, whereas age, smoking amount and baseline FEV1 were retained in the models. As shown in table 22,, exposure to endotoxin was associated with lung function declines in all TNF genotype groups. The estimated decline rates, however, were greater in the variant genotypes. Among the group with the TNF polymorphism, the average annual change in FEV1 associated with endotoxin exposure was –6.83 ml for G/A+A/A genotypes, in contrast to –2.87 for G/G. Among the LTA polymorphism groups, the annual changes were –3.97 and –3.58 ml for A/G, G/G, respectively, in contrast to –2.03 for the A/A genotype.
When TNF and LTA polymorphisms were combined based on the number of variant alleles—that is, heterozygous genotype (G/A of TNF and A/G of LTA) had one variant allele, and homozygous genotype (A/A of TNF and G/G of LTA) had two variant alleles—the estimated endotoxin exposure associated FEV1 declines relatively tended to be greater with increasing variant alleles (table 22).). The coefficients in variant genotypes were significantly different from zero, whereas this was not the case in variant alleles.
Furthermore, we added the factor of cigarette smoking to the combined TNF polymorphisms and endotoxin exposure, and examined their joint effects with multiple linear regression models. Again, other relevant factors were adjusted. As shown in table 33,, in subjects with either smoking or exposure to endotoxin, variant TNF genotype was associated with a greater FEV1 decline than corresponding wild TNF genotype. Subjects with both endotoxin exposure and tobacco smoking had greater annual declines in FEV1 than any other subgroups, but the annual decline was not greater in the variant genotype subgroup than in the wild genotype subgroup.
In this study, we examined a possible link between TNF polymorphisms and chronic lung function changes related to endotoxin exposure in a cohort of textile workers. Overall, the differences in annual FEV1 declines were greater in exposed subjects with variant genotypes. The magnitude of the effect modification of TNF polymorphisms on lung function changes associated with endotoxin exposure ranged, on average, from 2–4 ml per annum, or 40–80 ml for 20 years, indicating that the effect modification was relatively small.
The difference in annual decline of FEV1 between variant and wild type TNF genotypes among those who were both exposed to endotoxin and smoked was barely detectable in the joint effects models (table 33).). There are at least two possible explanations for this. Firstly, the coexistence of both endotoxin exposure and cigarette smoking may have exerted a too strong chronic effect on airways that overwhelmed or masked the modest effect modification of TNF polymorphisms. Another possibility is that the small number of subjects with the variant genotype in this category may have resulted in inadequate study power to detect the effect modification. The former is more likely, given that the sample sizes in the other cells of the table were similar.
TNF has strong proinflammatory and immunomodulatory properties. However, it has been reported that TNF is associated with increased airway responsiveness in normal subjects, and with airway inflammation with a neutrophil infiltration.6 Although both of the variant genotypes of TNF and LTA correspond to higher baseline and inducible TNF expression levels,9,10 the underlying effects of TNF polymorphisms alone on longitudinal changes in lung function may not be strong enough to be detected, without consideration of their interaction with environmental factors. TNF has been found to be central to smoke‐induced acute inflammation and resulting connective tissue breakdown, and occurrence of emphysema.27 Exposure to endotoxin may result in higher expressions of TNF.22,23 Therefore, the effect of TNF polymorphisms can be observed when exposure to endotoxin or cigarette smoking exists, in which TNF polymorphisms modify the effect of these environmental exposures on respiratory outcomes.
Among several studies linking TNF polymorphisms to the development or prognosis of COPD, an Irish study suggested that the A/A genotype of the TNF polymorphism was predisposed to more severe airflow obstruction and a worse prognosis in COPD.17 A Japanese study suggested that the TNF polymorphism was significantly associated with smoking‐related COPD,18 and might be partly associated with the extent of emphysematous changes in patients with COPD.19 Among the studies focusing on occupational populations, a study conducted in China found that the TNF polymorphism was associated with coal workers' pneumoconiosis.28 A French study also reported that interactions of TNF and LTA with environmental exposure and intermediate response phenotypes were important components in the pathogenesis of coal workers' pneumoconiosis.29 The current study provides additional information on the connection between TNF polymorphisms and chronic changes in lung function in endotoxin‐exposed workers.
We acknowledge several limitations in this study. Firstly, this is a study with a moderate sample size. Although we observed effect modification of TNF polymorphisms on endotoxin‐related lung function decline, gene–environment interaction was not clearly seen in the stratification analysis, quite possibly because of insufficient study power. Secondly, the subjects who donated blood samples accounted for only 55% subjects of the original cohort. To identify whether the subjects being analysed differed from those who were not included in this study, we compared the demographic features and lung function testing data at baseline and follow‐up between the participants and non‐participants in this study. The results showed no significant differences between the two compared groups, indicating that the loss to follow‐up was non‐differential. Therefore, the results based on the subjects were unlikely to be substantially biased. Lastly, we evaluated only two polymorphisms of TNF that are related to TNF expression levels. It is possible that other polymorphisms of these two genes—for example, ‐376G/A, ‐238G/A and 489G/A30—or polymorphisms of other cytokine genes may modify the association between environmental exposures and respiratory outcomes. Further studies with a larger sample size are needed to address this issue, in which possible gene–gene interaction or gene–gene joint effects can thus be examined.
In conclusion, TNF polymorphisms may modify longitudinal changes in lung function associated with occupational exposure to endotoxin, in which variant TNF genotypes (A allele of the TNF and G allele of the LTA polymorphism) boost lung function declines. The results highlight the importance of further investigating the roles of host factors in the development of respiratory diseases in occupationally‐exposed populations.
The authors thank the members of the Shanghai field team; Professors Gu Xue‐qi, Lu Pei‐lian, and Ye Ting‐ting from Shanghai Medical University; the Shanghai Textile Bureau; workers and members of the First and Second Textile Mills and the First Silk Mill of Shanghai, China; Janna Frelich for data management, and Dr Stephen Olenchock of the National Institute for Occupational Safety and Health for endotoxin analyses.
COPD - chronic obstructive pulmonary disease
PL‐EM - Partition Ligation‐Expectation Maximization
TNF - tumour necrosis factor
The Study was supported by NIOSH Grant R01OH02421, NIH Fogarty International Institute grant TW00828, and Flight Attendants Medical Research Institute Young Clinical Scientist Award.